Evolutionary and new growth theories. Are they converging?
Transcript of Evolutionary and new growth theories. Are they converging?
Munich Personal RePEc Archive
Evolutionary and new growth theories:
are they converging?
Castellacci, Fulvio
2006
Online at https://mpra.ub.uni-muenchen.de/27602/
MPRA Paper No. 27602, posted 21 Dec 2010 13:32 UTC
Evolutionary and new growth theories:
Are they converging?
Fulvio Castellacci
Department of International Economics, Norwegian Institute of International Affairs (NUPI)
POB 8159, Dep. 0033 Oslo, Norway Tel.: +47-22994034 Fax: +47-22362182
E-mail address: [email protected] (F. Castellacci)
This version: January 2006
Abstract
The article presents a critical review of evolutionary and new growth theories. The
purpose is to discuss the often-made claim that the two approaches, both inspired by
Schumpeter’s seminal work, are becoming more and more similar in terms of the
sources and mechanisms of the growth process on which they focus. According to this
argument, some kind of theoretical convergence between the two paradigms is taking
place. Differently from previous surveys of the field, the article compares evolutionary
and new growth theories by focusing on their major theoretical foundations. The
discussion leads to the conclusion that the two approaches greatly differ with respect to
all of their main theoretical building blocks, and that no convergence between the two
paradigms is therefore taking place. This finding should be welcomed by both
evolutionary and new growth scholars, because it is the process of interaction and the
fruitful exchange of ideas between different approaches that lead to advances in growth
theory, not their convergence to a common paradigm.
Key words: Innovation; economic growth; evolutionary economics; new growth theory.
1. Introduction
The crucial role of innovation for economic growth has been increasingly recognised in
the last two decades. Taking inspiration from the works of Schumpeter (1934, 1939 and
1943), a surge of interest in the study of innovation and growth started at the beginning
of 1980s with the seminal contributions in modern evolutionary economics (Dosi, 1982;
Freeman et al., 1982; Nelson and Winter, 1982). The new wave of theorizing was
motivated by the unsatisfaction with the stylized view of technological change
presented by the Solow model (1956). Sharing a similar criticism, new growth theorists
made a great effort to refine the Solovian view by building up models of innovation-
driven endogenous growth (Romer, 1986; Lucas, 1988; Aghion and Howitt, 1992).
Evolutionary and new growth theories have rapidly developed in the last two decades.
The great surge of interest in the new growth tradition, both in terms of formal
endogenous models and of the related econometric work, is well-known, and there
already exist various comprehensive surveys of the field (Temple, 1999; Islam, 2003;
George, Oxley and Carlaw, 2004). The development of evolutionary economics has also
been remarkable, and various critical discussions point to the strong similarities existing
between the different strands of research within the evolutionary paradigm (Andersen,
1994; Nelson, 1995; Archibugi and Michie, 1998; Nelson and Winter, 2002).
Both evolutionary and new growth scholars have repeatedly recognised Schumpeter’s
work as a major source of inspiration. Evolutionary scholars have frequently pointed to
the strong connections between Schumpeter and modern evolutionary economics
(Hodgson, 1997; Fagerberg, 2003). New growth theorists have also been invoking
Schumpeter as their main source of inspiration. Endogenous growth models have
increasingly incorporated some of the Schumpeterian ideas on the process of
technological competition and innovation-based growth, with the consequence of
1
making the outcomes of the new growth models closer and closer to the ones of
evolutionary models (Aghion and Howitt, 1998).
This leads to some major questions: given that the Schumpeterian insights on the
process of economic development constitute the main source of inspiration for both
evolutionary and new growth theories, how similar are the two approaches? Can we
observe theoretical convergence between the two, as it is often argued (Heertje, 1993;
Romer, 1994; Ruttan, 1997; Sarkar, 1998) on the basis of the Schumpeterian flavour of
both theories? These are the questions that this paper intends to answer. The paper
carries out a critical survey of evolutionary and new growth theories with the purpose of
analysing whether some kind of theoretical convergence is taking place between the two
paradigms.
The paper differs from previous critical surveys of the field in two main respects. First,
the comparison between evolutionary and new growth theories will not simply be
carried out by focusing on the formal growth models developed in the two traditions,
but will also include other strands of empirically-oriented and non-formal studies. The
latter constitute, in fact, an increasingly important part of growth theorizing, and
provide inspiration and new insights for the development of modelling exercises. By
enlarging the scope of the comparison, the paper will argue, it is possible to shed new
ligths on the similarities and differences between the two theoretical paradigms.
Secondly, the comparison will not be made in terms of the properties and results of
evolutionary and new growth models, but it will analyse, at a more general level, the
theoretical foundations of the two paradigms. By theoretical foundations we mean the
theoretical characteristics that may be defined as the major building blocks of growth
theorizing. The paper will consider six main theoretical foundations, and analyse them
by discussing the following questions:
2
(1) What is the main level of aggregation on which the theory focuses?
(2) Is it based on the notion of representative agent or on the one of heterogeneous
individuals?
(3) What is the mechanism of creation of innovation and new variety?
(4) What is the dynamics of the growth process? How is history conceived?
(5) Is the growth process deterministic or unpredictable?
(6) Does economic growth tend towards equilibrium, or is it a never ending process?
Section 2 will present these six questions in further details, and it will define the main
concepts used in the survey. Section 3 will use such theoretical questions as a
framework to discuss the basic characteristics of the different streams of evolutionary
economics developed in the last two decades, namely the neo-Schumpeterian long wave
theory, the technology-gap approach, Nelson and Winter-like evolutionary theorizing,
and the national innovation systems framework.1 In these evolutionary strands of
theory, technological change is the main engine of economic growth, which is regarded
as a complex process of transformation and qualitative change.2 The section will
suggest that, to a large extent, these approaches share the same theoretical foundations,
and may then be regarded as different strands of research within the same evolutionary
paradigm. It will also point, though, to the existing tensions and formidable challenges
currently faced by evolutionary economics. Section 4 will consider new growth theory
by briefly looking at both, formal models and econometric works, and by discussing
their theoretical foundations. Finally, section 5 will conclude the survey by pointing out
the great differences still existing between evolutionary and new growth theories, and
by claiming that no theoretical convergence is taking place between the two paradigms.
3
2. The theoretical foundations for the study of innovation and growth
This section presents the six theoretical questions that we will use to discuss and to
compare evolutionary and new growth theories in the remaining of the paper. We
believe that these six aspects constitute the major theoretical foundations for the study
of innovation and growth, and thus represent the relevant characteristics that it is
necessary to look at in order to compare different approaches. The section defines the
main concepts used in the survey, and it briefly points to the origin of each concept in
the history of economic thought. The reference to classical authors (e.g. Smith, Marx,
Veblen, Schumpeter) will be brief and stylized, as the purpose is not to carry out an
articulated discussion of the theoretical origins of modern theories of innovation and
growth, but rather to define some important concepts and to introduce the analysis to be
developed in the following sections.
2.1 What is the main level of aggregation?
A first important distinctive feature of theories of innovation and economic growth is
the level of aggregation chosen as fundamental starting point to build up the theory.
Three major positions may be distinguished in the history of economic thought:3
(i) Methodological individualism: this is the approach typical of classical and
neoclassical economists, as well as Schumpeter.4 According to this, the aggregate
properties of the economy must be studied by starting from the analysis of the
microeconomic behavior of consumers and firms. The whole economic system must be
analysed by looking at its component parts, the macroeconomic theory must necessarily
be microfounded.
4
(ii) Methodological holism (Hodgson, 1993, p.238): this is the approach typical of Karl
Marx, and later frequently adopted in economic sociology and heterodox
macroeconomics. 5 Here, it is the social and macroeconomic structure to determine the
behaviour of economic agents. The component parts of the economic system can only
be studied by analysing the whole, the microeconomic element depends to a large extent
on the macroeconomic structure.
(iii) Non-reductionism: the previous two positions are both said to be ‘reductionist’, in
that they only consider a one-way relationship between different levels of aggregation:
either the micro determines the macroeconomic element, or the latter affects the former.
An alternative to these reductionist views, not fully developed yet in modern economic
theory, was proposed long ago by Veblen (1899 and 1919).6 He suggested that
important interrelationships exist between the formation of individuals’ habits of
thought and aggregate institutional regularities. In his view, the macroeconomic and
social regularities are determined by the behaviour of individuals, but economic agents
are in turn greatly affected by the macro structure in which they live. This temptative
description of a co-evolution between different levels of analysis may be labelled ‘non-
reductionism’ (Hodgson, 1993, pp. 246-248), in that there does not exist a single
dominant level of aggregation, but each level interacts with the others. As section 3 will
point out, modern evolutionary economists frequently call for some form of non-
reductionism in the attempt to analyse the co-evolution across different levels of
analysis (Dosi and Winter, 2000), although no significant advance in this respect has
been obtained yet since the times of Veblen.
5
2.2 Representative agent or heterogeneous individuals?
This question refers to the way in which (micro) economic agents are represented in the
theoretical framework. In the history of economic thought, we may distinguish between
some major different approaches:
(i) Neoclassical typological thinking: this is the position adopted by neoclassical
economics since the marginalist revolution, according to which economic agents can be
studied by analysing the behaviour of a ‘representative agent’. In the simplest and most
standard version of the neoclassical metaphor, the economic agent is typically described
as a rational maximizer of utility/profits under given constraints and perfect
information.
(ii) Smithian typological thinking: according to Adam Smith and, later, Herbert Spencer
economic agents are genetically similar and homogeneous, but the production process
and the division of labour bring differentiation in skills and tasks because individuals
learn ‘by doing’. Heterogeneity, in this case, is not a precondition but a consequence of
the process of economic growth. It is not a genetic attribute of economic agents, but a
characteristic acquired during their working life.
(iii) Marxist typological thinking: Karl Marx pointed to the existence of a fundamental
opposition between two different social classes, the capitalists and the proletarians.
These two classes are defined in terms of their relationship to the means of production,
and have permanently different interests and purposes. However, within each class,
individuals are homogeneous. In other words, Marx implicitly assumes the existence of
a duality between social classes at the macroeconomic level, but not heterogeneity of
individuals at the microeconomic.
(iv) Schumpeterian typological thinking: in Schumpeter (1934 and 1939), the
microeconomic description of economic agents is rather peculiar. On the one hand,
6
there is a group of individuals, the entrepreneurs, genetically endowed with special
psychological characteristics; they are the ones to determine the growth process, the real
source of change. On the other hand, however, all the other economic agents are
ordinary and indifferentiated individuals, not dissimilar from the representative agents
of neoclassical theory, which react in a deterministic way to the changes of the process
over time. This appears as an intermediate position between neoclassical typological
thinking and evolutionary biology’s ‘population thinking’.
(v) Veblen’s population thinking: heterogeneity of economic agents is an essential
characteristic of an evolutionary approach to economic change, an early example being
the work of Veblen (1899). Inspired by the developments of evolutionary biology,
Veblen believed in the existence of a fundamental element of heterogeneity in
individuals’ cognitive processes and in the formation of habits of thought, and originally
pointed out that this variety is an important precondition for the process of economic
growth and social change. Applying the biological metaphor to economics, ‘population
thinking’ means that economic theory cannot ignore the heterogeneity in the population
of economic agents, but must necessarily be built on that, as variety is the major source
of novelty (innovation) in the process of economic development.7
2.3 What is the mechanism of creation of innovation and new variety?
In modern theories of innovation and growth, the mechanism through which innovation
and new variety are introduced in the economic system is the main source of economic
growth, and it thus constitutes a key element in the theoretical framework. Referring to
classical authors, we may briefly outline these different mechanisms of technical
change:8
7
(i) Manna from heaven: The neoclassical representative firm, in its simplest description,
is assumed to have perfect and complete knowledge about the best technology available
at any given time, and to always be able to adopt it. Technological knowledge is static,
perfectly codifiable, and independent of the economic context and situation in which
firms make their technological choices. All firms, then, can easily imitate and adopt
advanced techniques used by more innovative firms. Knowledge is regarded as a public
good, promptly available to all economic agents without further constraints. Technical
change, in the most simplified version of the neoclassical metaphor, is exogenous and
unexplained.
(ii) Learning by doing: this is the mechanism originally suggested by Adam Smith. The
production process brings deeper division of labour and increasing specialization, and
economic agents learn ‘by doing’ things and by producing goods during their working
activities. Innovation, being a necessary consequence of the productive process, is
therefore endogeneous, and mainly incremental and continuous.
(iii) Labour saving technical change: this is the mechanism pointed out by Marx,
according to which capitalists introduce labour saving technical innovations to decrease
labour costs and to expand their profits. However, it remains unclear in the Marxian
view how the new technology is invented, selected and adopted by capitalists.9 The real
mechanism of technical change is then exogenous and unexplained.
(iv) Schumpeterian innovation: Schumpeter was the first author to use a broad concept
of innovation which encompassed technical as well as organizational changes, and to
give it a central role in the explanation of economic development. Focusing on radical
rather than incremental innovations, he put forward the idea that ‘new combinations’ are
introduced by the entrepreneurs, which are individuals endowed with special
psychological traits and creativity (Schumpeter, 1934). Later in his life, he suggested
8
that the innovative process is systematically organized and performed by R&D
laboratories within large firms (Schumpeter, 1943), rather than introduced by creative
entrepreneurs.
(v) Veblen’s idle curiosity: Veblen suggests the existence of an important source of
variety which continuously opposes to the inertial nature of habits of thought and
institutions, namely the “human tendency towards experimentation and creative
innovation” (Hodgson, 1993, p.127). Veblen called this tendency ‘idle curiosity’, and
regarded it as a genetic human attitude that is a pre-condition for the process of growth,
and not a consequence of it (as in Smith and Spencer). Veblen conceived ‘idle curiosity’
as analogous to mutations in Darwinian evolutionary biology, and thus as an ongoing
and permanent source of change and renewed variety in the economic system.
2.4 What is the dynamics of the growth process? How is history conceived?
This characteristic refers to the type of dynamics of the growth process, and it is closely
related to the way in which history is conceived in the theoretical framework. A brief
look at the history of economic thought suggets that we may distinguish between some
major different ways of conceiving economic dynamics:
(i) Transitional dynamics: the focus of neoclassical theory is on the static allocation of
resources at a given time, and dynamic analysis is conceived as an extension of the
equilibrium metaphor to the long run. Economic dynamics is regarded as a process of
transition towards a new state of equilibrium. History may therefore be thought of as a
process of uniform-speed transitional dynamics towards long run equilibrium, rather
than a process of irreversible and qualitative change.
(ii) Increasing complexity: this is the position adopted by Adam Smith and Herbert
Spencer, which were both deeply interested in issues of transformation and dynamics,
9
rather than resource allocation in a static context. They both argued that socio-economic
change proceeds towards an increasing degree of specialization and complexity, and
that it is a process of qualitative change. History was then conceived as a (uniform-
speed) gradual evolution towards higher states of complexity and differentiation.
(iii) Revolutionary and dialectic dynamics: Marx conceived the dynamics of economic
and social change as revolutionary, violent and disruptive. Growth is not a slow process
of incremental and continuous change, but rather a discontinuous and radical jump from
one stage of development to a better one. In his view, history may be conceived as the
succession of different phases that proceed in a dialectic and revolutionary manner, until
the final state of rest, communism, ultimately sets in.
(iv) Saltationist dynamics: Schumpeter argued that “social phenomena constitute a
unique process in historic time, and incessant and irreversible change is their most
obvious characteristic” (Schumpeter, 1954, p.435). Such a definition of evolution points
to the historical dependent unicity and irreversibility of the process of change, which is
meant to be qualitative as well as quantitative change. According to him, evolution may
be thought of “more like a series of explosions than a gentle, though incessant,
transformation” (Schumpeter, 1939, p.102). This ‘saltationist’ characterization of the
process of economic evolution is in many respects similar to Marx, and it is in sharp
contrast with the more ‘gradualist’ character of other classical economists as well as
neoclassical economics.
(v) Gradualist evolutionary dynamics: gradual, continuous and incremental qualitative
change is not only the characterizing element of Smith and Spencer, but also of the
evolutionary theory of Veblen. According to him (Veblen, 1899 and 1919), the
coexistence of forces driving towards change (‘idle curiosity’) and inertial forces (the
persistence of ‘habits of thought’ and ‘institutions’) determines a process of gradual
10
evolution. History is an evolutionary process of qualitative change and cumulative
causation.
2.5 Is the growth process deterministic or unpredictable?
Another important feature in economic growth theorizing is whether the process
described is deterministic and predictable, or rather non-deterministic and
unpredictable. Although many intermediate positions could be discussed, it is useful to
point out the two major (opposite) views.
(i) Mechanistic, deterministic and predictable process: the economic world is
understood and represented in terms of cause-effect mechanisms, in which there is no
space for purposeful behaviour and free choice. Inspired by the developments of
classical physics and astronomy, the mechanistic view in economics has been dominant
since the time of classical economists (including Marx), marginalist and neoclassical
economics. The mechanistic view implies that, given the initial conditions at the present
time and the law of motion of the economic system, any future state can be perfectly
foreseen. Mechanicism, therefore, implies determinism and predictability of future
economic outcomes.10
(ii) Non-mechanistic, non-deterministic and unpredictable process: the mechanistic
metaphor, according to a different view, is not appropriate to describe the evolution of a
complex system. Purposeful behaviour, deliberate choice and creativity of individuals
introduce a fundamental element of non-mechanicism and unpredictability in the
economic world. This is the view adopted, more or less explicitly, by German
Historicists, Old American Institutionalists (e.g. Veblen), and to a certain extent by
Schumpeter. In this view, the process of innovation and economic growth is
characterized by genuine and pervasive uncertainty, rather than mere computable risk.
11
This distinction was originally put forward by Knight (1921), according to which “the
practical difference between the two categories, risk and uncertainty, is that in the
former the distribution of the outcome in a group of instances is known (either through
calculation a priori or from statistics of past experience), while in the case of
uncertainty this is not true, the reason being in general that it is impossible to form a
group of instances, because the situation dealt with is in a high degree unique” (Knight,
1921, III.VIII.2). In an economic world characterized by radical uncertainty, rather than
computable risk, given the initial conditions at the present state, it is not possible to
predict with certainty what the future state of the economic system will be. The
economic process is non-deterministic and fundamentally unpredictable.
2.6 Towards equilibrium or never ending?
Where does the economic process lead to? Does it tend towards a final state of long run
equilibrium, or does it change continuously and go on moving forever without any
definite final point? By and large, it is possible to point out two main different views on
this fundamental characteristic of economic theory:
(i) Towards equilibrium: the process of economic growth tends towards a final state of
rest, equilibrium, and greater economic welfare. This was the view adopted, in different
forms, by Adam Smith and Herbert Spencer, as well as Marx (for which ‘communism’
is a final state of rest in which all conflicts and dualisms ultimately cease). The
equilibrium view became more explicitly dominant in economics after the marginalist
revolution, since the last decades of the nineteenth century. More recently, the
neoclassical theory of growth (e.g. Solow, 1956) extended the static concept of
equilibrium to the analysis of the dynamics of the long period, by assuming the
12
existence of a ‘steady state’ towards which the economic system will tend in the long
run.
(ii) Never ending process: the equilibrium view has frequently been criticized outside of
the economic mainstream. Economic growth, it has been argued, is a never ending and
ever changing process, it does not tend towards a steady state of balanced growth. This
is the view expressed in the past, among others, by German Historicists, Schumpeter
and Veblen. Using the latter’s words, economic evolution is “a continuity of cause and
effect. It is a scheme of blindly cumulative causation, in which there is no trend, no final
term, no consummation […], a theory of the process of consecutive change, realized to
be self-continuing or self-propagating and to have no final term” (Veblen, 1919, pp. 36-
37).
3. Evolutionary growth theorizing
After having defined the main concepts that will be used in this survey, we will now
discuss the major strands of research within modern evolutionary economics. This
section will consider in turn the main approaches, namely the neo-Schumpeterian long
wave theory, the technology-gap approach, Nelson and Winter-like evolutionary
theorizing, and the National Innovation Systems framework. Each subsection is
composed of two parts: the first presents a brief overview of the approach, while the
second part analyses its theoretical foundations by answering the six questions
presented in section 2. The discussion will point out that these four approaches share, to
a large extent, the same theoretical foundations, so that they may be conceived as
different strands of research within the evolutionary economic paradigm. The analysis
13
will also argue, though, that although important advances have been realized in
evolutionary economics in the last two decades, there still exist great challenges ahead.
3.1 Neo-Schumpeterian long wave theory
The neo-Schumpeterian approach to economic growth takes great inspiration from
Shumpeter’s book Business Cycles (1939), in which the author put forward a theory
about the existence of long waves of economic growth. His original point was to focus
on the importance of basic (radical) innovations in creating such long waves, because,
he argued, they have potentially a deep impact on the whole economy. The
Schumpeterian insights on the central role of radical innovations in the macroeconomic
growth process did not affect significantly the development of economic thought in the
following four decades. Since the mid 1970s, however, there started to be greater
criticism on the way in which mainstream economics approached the relationships
between technical change and economic growth, and a renewed interest in the central
role of innovation as the majour source of economic growth.
The debate started with Kuznets' (1940) review of Schumpeter's Business Cycles
(1939). His long wave theory, Kuznets argued, did not explain neither the reasons for
the timing of occurrence of basic innovations in the depression phase of the wave, nor
why they tend to cluster over time. Mensch (1979) put forward the idea that radical
innovations tend to cluster in the depression phase of the long wave because this is the
time in which the lag between invention and innovation is shortened (so-called
‘depression-trigger hypothesis’). A rich empirically oriented literature (among others
Kleinknecht, 1981; Van Dujin, 1983) focused on the timing of clustering of basic
innovations. The empirical results of these works have been heavily debated. On the
whole, as pointed out by Freeman, Clark and Soete (1982), the empirical evidence on
14
the clustering of basic innovations in the depression phase of the wave is rather weak
and not conclusive.
After this empirical debate, a second stream of neo-Schumpterian literature flourished
during the 1980s, providing a number of concepts and ideas useful to give a stronger
theoretical foundation to long wave theory. These more conceptually oriented
contributions started with the publication of the book Unemployment and Technical
Innovation (Freeman, et al., 1982), and was followed by the works of Freeman (1983,
1984 and 1987), Perez (1983, 1985), and Freeman and Louca (2001).
As Perez (1983) points out, the Schumpeterian process of development “unfolds within
the economic sphere conceived as a self-regulating organism which provokes its own
disturbances (innovations) and absorbs its impacts by constantly striving towards new
higher equilibria”. The social conditions and institutional framework are excluded from
the causation mechanism that drives the primary cycle. This is the reason why she
argues that “Schumpeter does lay the foundations for a theory of the cyclical nature of
the capitalist economy but not of long waves” (Perez, 1983, p.359). Based on these
considerations, neo-Schumpeterian scholars conceive the capitalist system as formed by
two related sub-systems: the techno-economic and the socio-institutional. It is the joint
evolution of these sub-systems to determine the ‘mode of development’, and
consequently the rise and fall of long waves.
According to this view, it is not important when a set of basic innovations is introduced,
but rather that these radical innovations are strictly interrelated and pervasive, i.e. that
they may drive the growth of many fast growing sectors of the economy. Such a family
of interrelated basic innovations may be called ‘technological system’ (Freeman et al.,
1982), ‘technological paradigm’ (Dosi, 1982), or ‘technological style’ (Perez, 1983).
This concept is arguably quite similar to that of ‘general purpose technologies’,
15
although the latter is more frequently used in new growth theories (see section 4). When
a new technological style arises, there is a big impulse in the techno-economic sub-
system to adopt the new best practice technology with high profit prospects. However,
the techno-economic system is more ready to accept and adopt changes, while the socio-
institutional one may take a longer time before making the changes required by the new
technological style. The mismatch between the two sub-systems may retard the large-
scale introduction of the new paradigm, precisely because social, organizational and
institutional changes are necessary before it can diffuse to the whole economy. As the
socio-institutional system evolves, the ‘harmonic complementarity’ between the two
systems gradually restores, and a new mode of development eventually sets in. This
may determine a long wave pattern similar to the primary cycle described by
Schumpeter (1939): rapid diffusion of the new paradigm, incremental innovations over
its ‘natural trajectory’ (Nelson and Winter, 1977), creative destruction, and
consequently the upswing and prosperity phases of the long wave. Later on, increased
competition and market saturation, decreasing revenues from the new technologies and
decline of profits, which characterize the recession and depression phases of the long
wave.
3.1.1 Discussion
The first question that our critical review considers refers to the level of aggregation of
the approach. In this respect, neo-Schumpeterian long wave theory is a macroeconomic
approach to the study of innovation and growth, the focus of the analysis being the
evolution of a country (or a group of countries) over time. Great attention is devoted to
the study of sectoral differences, focusing in particular on more technologically
advanced and fast-growing sectors, which are those that drive the overall growth of the
16
economy. The sectoral analysis, though, is primarily carried out with the purpose of
understanding the implications and effects of sectoral patterns on national and
international macroeconomic growth. Differently from Schumpeter, then, neo-
Schumpeterian theory is not explicitly microfounded. It shows that the main features of
Schumpeterian macroeconomics can be obtained without necessarily following
methodological individualism.
Consequently, as there is no description of the microeconomic level, the notions of
heterogeneity and population thinking are not explicitly considered in this approach.
However, the fundamental role of heterogeneity and, more in general, the evolutionary
foundation of such an approach, are increasingly recognized by recent long wave
studies (Freeman and Louca, 2001). An evolutionary type of modelling in which the
interactions of heterogeneous agents determine long wave patterns has already been
proposed in the works of Iwai (1984) and Silverberg and Lehnert (1994). The future
extension of this class of models could make the evolutionary foundation of neo-
Schumpeterian studies more explicit than it is at the present stage.
Similarly to all the other theoretical frameworks considered in this paper, innovation is
the main source of economic growth. The historical and institutional context in which
technical and organizational innovations take place is considered with great accuracy in
neo-Shumpeterian works. On the one hand, the innovative process is exogenous,
because it depends on the science and technology system, which is pointed out as
important but not explicitly investigated. On the other hand, though, innovation is an
endogenous activity, determined by R&D investments of firms and, in a later phase of
the long wave, linked to demand and production growth through learning by doing,
dynamic economies of scale, and embodied technical progress.
17
Innovation is arguably the majour source of economic growth in this framework, but
this does not justify the often-made claim that neo-Schumpeterian long wave theory is a
technological deterministic approach. Such a criticism is based on the fact that when a
new technological paradigm emerges, it is the evolution of the techno-economic system
to determine the socio-institutional characteristics that are required to compete in the
new long wave period. So, transformations in the techno-economic system affect greatly
the characteristics of the new mode of development. However, in the downswing phase
of the long wave, innovations are more likely to be introduced in the market because
firms and consumers are more willing to risk and to try out new solutions. It is in the
downswing phase that consumers’ expectations, firms’ animal spirits, and social and
political changes facilitate the introduction and diffusion of a new technological
paradigm. Therefore, changes in the socio-institutional system may also affect the
techno-economic, so that it is not appropriate to argue that neo-Schumpeterian long
wave theory is based on a simple one-sided and technological deterministic view of the
process of economic change.
The creation and diffusion of interrelated innovations determine long waves of
economic growth, each characterized by an initial speed up (upswing) and then a
slowing down phase (downswing). The dynamics of the process is saltationist,
disruptive, irregular, and characterized by structural and irreversible change, as in
Schumpeter’s view. Precise regularity and strict periodicity are not assumed in long
wave theory, the process repeats itself over time but in a rather irregular way (Freeman
et al., 1982). According to this interpretation of history, the recurrence of long waves
does not imply that the waves are all the same. The only recurrent mechanism is the co-
evolution between technological and socio-institutional changes and its importance for
economic growth, but the precise form that they take in each historical phase is ever
18
changing and always different. Every occurrence is singular and unique in historical
time.
A common criticism made to the long wave approach refers to its ‘mechanistic’ flavour.
On the one hand, it is true that, once a new technological paradigm emerges, the long
wave process is assumed to follow in a more or less automatic and mechanistic way,
closely resembling the Schumpeterian primary cycle. On the other hand, however, in the
downswing phase it is not possible to predict which technological and organisational
innovations will characterize the following historical phase, and when they will come
about. The outcomes of the science and technology system are non predictable with
accuracy, and the same is true for the socio-institutional changes that will follow.
Considering the whole long wave sequence, then, the process described may certainly
be regarded as non-deterministic and non-predictable.
Finally, with respect to the sixth theoretical question that we consider in our discussion,
it should be observed that the neo-Schumpeterian process of growth is ever-changing
and never-ending, it does not tend towards the steady state. Similarly to Schumpeter, the
economic system is never in equilibrium, there are always forces determining further
disequilibrating movements. It is innovation that continuously breaks the circular flow
of economic activity, and that determines the inherent disequilibrium nature of the
economic system.
3.2 The technology-gap approach
While neo-Schumpeterian scholars study the process of economic development within
each country, technology-gap theorists focus on technological and economic differences
between countries. The approach has originated from the contributions of historically
oriented economists, which investigated the process of catching up and overtaking of
19
some advanced (leader) countries in the last two centuries by focusing on the creation of
new technologies and on its international diffusion (Veblen, 1915; Gerschenkron, 1962;
Habakkuk, 1962; Landes, 1969; Abramovitz, 1986 and 1994, Freeman, 1987). These
historical contributions, different as they may be, all point out that two broad sets of
factors are necessary for succesful catching-up and rapid growth in the long run: techno-
economic and socio-institutional factors. The crucial point is thus that catching up is a
complex process, so that its investigation cannot only look at economic factors, but also
at the important technological, social and institutional aspects related to the
development process.11
Originating from these historically oriented studies, a modern strand of technology-gap
theory has developed since the 1980s. These more quantitative-oriented applied studies
aim at explaining the historical evidence on catching up by adopting a Schumpeterian
perspective on the importance of innovation and international diffusion for economic
growth. The Schumpeterian idea that firms compete in the market by upgrading their
technological capabilities is then applied to the macroeconomic level, where countries
are assumed to compete for the economic leadership through their technological
capabilities, absorptive capacities and innovating activities. Econometric works in this
tradition typically investigate differences in economic growth rates and trade
performances by using indicators of national technological activities, such as R&D and
patent statistics. The strong correlation generally found between technological and
macroeconomic performance (e.g. Fagerberg, 1987 and 1988; Dosi, Pavitt and Soete,
1990) is then taken as an indication of the fundamental role played by the creation and
diffusion of technologically advanced products and processes for explaining growth rate
differences.
20
These econometric studies investigate differences in technological and economic
performances on large samples of advanced and middle-income countries, so that,
compared to the previous historically oriented contributions, the focus shifts from the
study of the catching up process of single countries to the analysis of convergence and
divergence in the whole sample of countries, carried out through statistical and
econometric techniques. The cross-country econometric methodology is thus
remarkably similar to that used in the convergence literature in mainstream economics
(see section 4).
The theoretical perspective that underlies the applied work in the evolutionary
technology-gap tradition, however, is quite different from its neoclassical counterpart.
As developed by Cornwall (1977), Abramovitz (1986 and 1994), Fagerberg (1987, 1988
and 1994) and Verspagen (1991 and 1993), the modern technology-gap approach to
economic growth assumes that innovation and the international diffusion of new
technologies are the main sources of differences in growth rates between countries.
Follower countries have a technology-gap (or technological distance) that separates
them from the leader country, and they can therefore try to exploit their backward
position by imitating and using advanced technologies developed by the leader country,
instead of creating them from scratch. The process of imitation and diffusion of new
technologies is costly, though, and it requires the existence of social and institutional
capabilities that not all the follower countries have (Archibugi and Michie, 1998). This
explains why catching up and convergence are not automatic and common outcomes.
Considering the conditions that are necessary for successful imitation and catching up,
two broad sets of factors have been stressed. First, following Abramovitz (1994, p.24),
it is important to consider the ‘technological congruence’ of a country. This is defined
by various factors: (i) the ‘technological interrelatedness’, i.e. how much a country is
21
committed to the old technological paradigm, and therefore how difficult is to make the
jump into the emerging one; (ii) the country’s natural resources and factors endowment;
(iii) consumers’ demand and tastes; (iv) market size and scale; (v) transportation and
infrastructure; (vi) facilities for structural change, i.e. how rapidly the economic system
is able to shift resources from the old to the new paradigm; (vii) general macroeconomic
conditions and the rate of growth of demand. Second, turning to the broad set of social,
cultural and institutional factors, Abramovitz (1994, p.25) defines the ‘social capability’
of a country, characterized by: (i) its level of education and technical competence; (ii)
skills of the entrepreneurial class; (iii) commercial, industrial and financial institutions;
(iv) political and social characteristics that influence the risks and incentives of
economic activity; (v) science-technology links in firms and public research centres.
Considering them together, techno-economic congruence and social capability differ
between countries in each technological paradigm, and these structural differences may
explain why some countries manage to succesfully catch up with the technological
leader, while some others fall behind. The major difference between this theoretical
perspective and technology-gap models in the neoclassical tradition (or North-South
models, see Chui et al., 2002) is that the evolutionary view stresses the importance of
the social and institutional structure to determine the social capability of a country
(Abramovitz, 1986) and its ability to imitate foreign technologies.
3.2.1 Discussion
Similarly to the neo-Schumpeterian long wave theory previously discussed, the
technology-gap approach is a macroeconomic approach. Applied works in this tradition
are not based on the concept of aggregate production function, and the approach is
therefore not explicitly microfounded, as there is no description of the behaviour of
22
economic agents that may determine the aggregate outcomes. The Schumpeterian
insights on innovation, diffusion and technological competition are transferred from the
individual to the aggregate level of analysis: such as firms compete in the market for
their market shares and profits, countries compete in the international arena for the
technological and economic leadership. The main unit of analysis, then, is the country
(some recent works focus on the regional level instead; see Fagerberg and Verspagen,
1996; Cappelen et al., 2003).
As the approach is not microfounded, heterogeneity of individuals and population
thinking are not explicitly considered in this framework. It is argued that countries are
fundamentally and structurally different, particularly from a social and institutional
point of view, but this variety at the macroeconomic level is assumed, and not explained
by focusing on the interactions in a population of heterogeneous agents. Some formal
evolutionary models (Dosi and Fabiani, 1994; Dosi et al., 1994), however, show that the
interactions of heterogeneous agents in an evolutionary framework may generate
situations of catching up and falling behind, and reproduce the empirical patterns of
convergence and divergence that applied studies have found. More work of this kind is
needed in the future. The study of the aggregate properties of microfounded
evolutionary models is a fascinating challenge for future research in this field, and it
may provide a bridge between the applied work in the technology-gap tradition with the
formal analysis of Nelson and Winter-like evolutionary models (discussed in section 3.3
below).
Technological change is the main source of economic growth, but its mechanism, rate
and direction are not investigated. Although innovation and diffusion of technologies
are conceived as fundamental conditions for catching up, the way in which they are
introduced in the economic system is not analysed further. The focus is on the structural
23
and institutional factors that may facilitate or hamper the process of international
diffusion, but not on the factors that may explain a differential rate of creation of
innovations in different countries. The approach does not shed any new light in this
respect.
The dynamics implicitly assumed in technology-gap studies is mainly saltationist. The
approach argues in fact that when a new technological paradigm sets in, there is a strong
technological push in the economic system, which may turn to have important
consequences for the patterns of convergence/divergence. In times of radical changes,
leader countries can more easily invest in the new technologies, and are therefore likely
to grow faster than follower countries, so that greater divergence between rich and poor
countries may follow. After some decades, when catching-up countries start to imitate
and use the new technologies in large scale, convergence in the whole sample may be a
more common result. So, the dynamics of convergence/divergence between countries
does not proceed at uniform speed, but it rather follows the paradigmatic, saltationist
and discontinuous character of technological change.
Moreover, the process of catching-up and falling behind is conceived as non-
deterministic and non-predictable. The reason is that as technological change is
fundamentally an uncertain phenomenon, it is not possible to predict which new
technological system will prevail in the future. Therefore, it is hard to predict with
accuracy the countries that will be more likely to catch-up in the future, and those that
will fall behind. The applied works in this tradition show the changing character of the
catching up and convergence process over time. The more recent evidence points out
that, while it was relatively easier to imitate and import foreign technologies in the age
of Fordism and mass production, the scope for catching up has significantly decreased
in the last two decades. The catching up process of follower countries requires now a
24
greater effort for the creation and improvement of national technological capabilities
(Fagerberg and Verspagen, 2002). The technology-gap process of growth is then ever-
changing and never-ending, it is a process of qualitative change and transformation,
rather than a transition towards the steady state.
3.3 Nelson and Winter-like evolutionary theorizing
The possible use of the biological-evolution metaphor in economic science was
originally suggested by Veblen and Marshall more than a century ago, but the
development of modern evolutionary economics is relatively recent, tracing back to
Nelson and Winter’s (1982) book An Evolutionary Theory of Economic Change. Nelson
and Winter-like evolutionary theorizing is currently the most influential and rapidly
developing branch in the evolutionary economic paradigm.
Three complementary streams of literature have recently extended in various directions
Nelson and Winter’s theory of economic change: (i) microeconomic evolutionary theory
of consumers, firms and organizations, closely connected to cognitive psychology,
business and organizational studies;12 (ii) sectoral studies on the historical evolution of
particular industries, and related analyses of industrial dynamics and sectoral systems of
innovation;13 (iii) formal models of economic growth.14 Although the three streams
focus on different aspects of the evolutionary process at various levels of aggregation
(firms, sectors and countries, respectively), they all conceive economic evolution as
driven by the interactions between heterogeneity, selection and innovation processes.
Figure 1 shows a simplified scheme of these interactions.
Heterogeneity (or variety) of economic agents is a fundamental feature of the
evolutionary economic world. The latter is characterized by complex evolving
knowledge, bounded rational agents and radical uncertainty. In such an uncertain world,
25
individuals follow routines and habits of thought in their economic activities. Routines
are regarded as the counterpart of genes in biological evolution. The reason for this
analogy is threefold: routines are embodied in the minds and production activities of
economic agents; they greatly differ among the various units of the population; and they
can be transmitted from one individual to another, so that they may take account of the
regularities sustaining stable and inertial patterns of production over time.
Within the same firm, production can be conceived as guided by routines at different
levels, driving the standard operating procedures, the investment behaviour, and the
deliberate search for new routines or solutions when the old ones prove to give
unsatisfactory results in terms of market shares and profits. Routine-guided firms may
thus be thought of as the counterpart of phenotypes in biological evolution, because
their behaviour is the result of the interactions of their genetic endowment (individual
skills and organizational routines) with a given economic and institutional environment.
Since Nelson and Winter (1982)’s seminal work, several evolutionary models of
economic growth have tried to formalize this idea of routine-guided heterogeneous
firms within a disequilibrium framework. These models assume that firms differ with
regards to the techniques that they use (Iwai, 1984; Conlisk, 1989; Silverberg and
Lehnert, 1994), their behaviours and strategies (Chiaromonte and Dosi, 1993; Dosi et
al., 1994; Fagiolo and Dosi, 2003; Silverberg and Verspagen, 1994a; 1994b; 1996), or
the characteristics of the sectors in which they operate (Winter, 1984; Verspagen, 1993).
Evolutionary analytical models, therefore, aim at reproducing the idea that the
‘routinized’ character of the productive process carried out by a population of
heterogeneous firms may generate a relatively stable pattern of economic activities and
relationships over time. The important point, however, is that such inertial forces and
inherent persistency are continuously counteracted by dynamic forces that push the
26
economic system towards evolution, change and transformation. These dynamic forces
are technological competition and selection, on the one hand, and innovation on the
other.
In the same way as animal species compete for their survival in the natural environment,
heterogeneous firms compete in the market by trying to employ more advanced
techniques, and to produce at lower costs and better quality than their competitors. The
selection mechanism in evolutionary models typically depends on the profits realized by
each firm. Firms that are able to obtain high profits increase their market shares; firms
with inferior technological capabilities realize lower profits, loose market shares, and
will ultimately be driven out of the market. The idea of selection-based growth, put
forward in different forms in the past by Schumpeter (1939), Alchian (1951) and Winter
(1964 and 1971), is usually represented in recent formal models through the use of
replicator (or Lotka-Volterra) equations in which the firm’s market share (or production
level) is assumed to evolve over time as a function of its technological capability and
profitability.
An important qualification, made by the growing number of studies of sectoral patterns
of innovation (Pavitt, 1984; Malerba, 2002), is that the competition-selection process
works differently in different industries of the economy. Each sector is characterized by
the complex interactions between heterogeneous agents, economic structure, institutions
and technological characteristics. The latter, in particular, determine the ‘technological
regime’ in which competition and selection take place. The technological regime may
be conceived as the technological environment in which innovative activities take place
in different industries of the economy. Such an environment differs in terms of
technological opportunities, properties of the knowledge base, cumulativeness and
appropriability conditions. Formal models and econometric evidence show that the
27
characteristics defining technological regimes may generate the different patterns of
industrial dynamics originally identified by Schumpeter (i.e. the so-called Schumpeter
Mark I and II; see Winter, 1984 and Malerba, 2005).
Over time, competition and selection tend to consume and to reduce the initial
heterogeneity. Without the creation of new variety, the process of evolution would soon
come to an end. The fundamental point about the evolutionary economic world is
precisely that there is an ongoing introduction of novelty, so that heterogeneity and
variety are continuously renewed, and evolution is a never-ending process. In particular,
two main different sources of novelty have been stressed in the literature. The first is a
kind of ‘unintended’ innovation, that arises when new routines are created as an
automatic and non-deliberate consequence of routinized production within firms. This is
for example the case when the firm expands its production scale by hiring additional
workers or buying new machines. The additional workers and equipments can never
exactly replicate what the old were doing, so that a firm’s routines can be randomly
modified at any time (Nelson and Winter, 1982, ch. 5). Moreover, the old routines
applied to a larger scale can be improved simply because workers learn by doing and by
producing. Dynamic economies of scale assume then an important role in an
evolutionary environment, as it is for example the case in the model by Silverberg, Dosi
and Orsenigo (1988).
A second important source of novelty comes from a deliberate search for new technical
solutions whenever the old one does not lead to efficient outcomes and satisficing
profits. Nelson and Winter (1982)’s formal model assume that when the profit rate falls
below a certain threshold, the firm will engage in a process of search for a better
technique by imitating other firms or by creating innovation. Winter (1984) and
Malerba (2002) point out that the probability that a firm chooses to imitate or to
28
innovate depends on the characteristics of the technological regime in which it operates,
and in particular on the possibility to appropriate the innovation profits, which
determines the technological spillovers that is possible to exploit in a given sector of the
economy. A later class of evolutionary models (Silverberg and Verspagen, 1994a;
1994b; 1995; 1996), has introduced the idea that firms may change their strategies and
routines by learning from past experience, so that evolution does not only imply
technological change but behavioural learning as well.
In a nutshell, evolutionary economic theory explains growth in terms of the dynamic
interactions between heterogeneity, competition, selection, and innovation, where the
latter leads to renewed heterogeneity and thus to perpetuate the growth process. From a
theoretical point of view, the evolutionary description of the economic world appears as
a novel contribution to growth theory. Its empirical relevance, though, is still difficult to
evaluate, and the relationship between formal models and econometric work in this
tradition has not been made explicit yet. In particular, what kind of empirical stylized
facts may be generated as outcomes of evolutionary models?
Although evolutionary economics has not yet agreed on a standard set of assumptions
and results, important empirical trends have been generated as ‘emergent properties’ of
different classes of evolutionary models, i.e.: (i) structural change and creative
destruction (like in the studies of industrial dynamics, history-friendly models and
recent studies on ‘sectoral systems of innovation’, see Malerba, 2005); (ii) path-
dependency (in models where the coexistence of random events and increasing returns
may generate path dependent phenomena of the kind described by David, 1985, and
Arthur, 1994); (iii) long waves and fluctuations without fixed periodicity (Silverberg
and Lehnert, 1994; Silverberg and Verspagen, 1994a; 1994b; 1995; 1996), reproducing
the predictions of the neo-Schumpeterian literature discussed in sec. 3.1; (iv)
29
endogenous specialization patterns and uneven international trade (e.g. Verspagen,
1993); (v) patterns of convergence/divergence between countries at the macroeconomic
level (Dosi et al., 1994; Chiaromonte and Dosi, 1993), closely related to the predictions
of the technology-gap approach discussed in sec. 3.2. The examples above indicate that
an important future challenge for Nelson and Winter-like evolutionary modelling is to
reach a closer link between formal models and econometric studies, as this may also
provide a more explicit bridge with the other evolutionary traditions considered in this
paper.
3.3.1 Discussion
Evolutionary models are explicitly microfounded on a population of heterogeneous
agents (population thinking). The theory is bottom-up built, it considers necessary to
start from the microeconomic level to derive the macroeconomic implications.
Aggregate phenomena are defined as ‘emergent properties’, i.e. “the collective and
largely unintentional outcome of far-from-equilibrium micro interactions” (Dosi and
Winter, 2000, p. 5). Individuals’ skills and firms’ routines are the basic units of
microeconomic analysis. However, the way in which routines and habits of thought of
economic agents may in turn be shaped and affected by the macroeconomic and
institutional environment in which they operate has not been made explicit yet. A co-
evolution across different levels of analysis is in fact often called for, but not
systematically introduced yet in evolutionary modelling. Arguably, future evolutionary
models may assume the same non-reductionist character as verbal and non-
mathematical studies of evolutionary economic change. Non-reductionism appears
therefore as an important challenge for future research in evolutionary economics.
30
The population of heterogeneous firms is analyzed in terms of the interactions between
variety, competition, selection and innovation. The latter is the fundamental source of
renewed variety and economic growth, without which evolution would ultimately cease.
Novelty can take the form of unintended and automatic consequence of the routinized
production, or of deliberate search by firms for new technologies (through innovation
and imitation). In both cases, the focus of evolutionary modelling is on an incremental
type of innovation, while Schumpeterian radical innovations have not yet found a
decisive role in this strand of research.
As a consequence, the dynamics of the economic system is prevailing gradualist,
characterized by slow and continuous change and transformation, rather than drastic
jumps as in the case of neo-Schumpeterian and technology-gap approaches. History
may be conceived as an evolutionary process of gradual and continuous growth and
qualitative change, which does not necessarily pass through the succession of different
paradigmatic phases. The focus on routinized production and routine-guided innovation
gives the theory a bias towards continuity and gradual evolution, in which radical
technical and institutional changes can hardly emerge from the inertial quality of
routines and habits of thought.
Similarly to the other evolutionary strands of research, economic growth is seen as a
non-predictable process, because fundamental sources of uncertainty exist in the
economic system. In particular, uncertain and non-deterministic innovative activity is
represented in formal models by assuming that the arrival rate of innovation follows a
stochastic process (e.g. a Poisson random variable). This formalization, though, appears
more suitable to represent an economic environment characterized by computable risk
rather than the strong and pervasive uncertainty of the evolutionary world.
31
The coexistence of random and systematic factors driving economic evolution (Nelson,
1995; Verspagen, 2005), together with the coexistence of inertial and dynamic forces,
determine the outcomes of the models: structural change, path-dependency, aggregate
growth, endogenous specialization patterns, and convergence/divergence across
countries. All of these phenomena are explained as the result of far-from-equilibrium
micro interactions. Differently from the neoclassical metaphor of a steady state, then,
evolutionary economics theorizes an ever-changing and never-ending process of growth
and transformation.
32
Figure 1. Main relationships in Nelson and Winter-like evolutionary economics
Technological knowledge: tacit, embodied,
interactive, context-dependent, cumulative
Bounded rationality: limited capabilities
and ‘satisficing behaviour’
Heterogeneity: different routines (genes), different
firms (phenotypes)
’Routinized’
production
Replication, inertia and stability
Competition and
selection Innovation
Outcomes: structural change; path dependency;
long waves; convergence/divergence
33
3.4 The national innovation systems framework
The previous sections have already made clear that the heart of evolutionary growth
theorizing is the investigation of the innovative process. Many feedbacks and complex
interactions are involved in the creation of technical and organizational innovations,
between individuals within the same firm, between different firms, between producers
and users of the new technology, between public and private organizations. Towards the
end of the 1980s, it was increasingly recognised that such complex links could not be
studied within a linear framework, and a ‘systemic’ approach to the study of innovation
was developed.
Freeman (1987) was the first to use a systemic perspective in his national case study on
Japan. The national innovation system (NIS) (defined as “the network of institutions in
the public and private sectors whose activities and interactions initiate, modify and
diffuse new technologies”, Freeman, 1987, p.1), he argued, is arguably the most
important factor behind the spectacular economic performance of Japan after World
War II. From his neo-Schumpeterian perspective, Freeman conceived the national
innovation system as a subset of the ‘socio-institutional’ one. The case of Japan showed
how important it is for a country to have an active and well-organized innovation
system.
Conceiving the innovation system in a more narrow way than Freeman, Porter (1990)
studied the factors behind the different economic performance of ten industrialized
countries. He argued that innovative activities greatly vary in firms of different
countries for the existence of differences in the following aspects: (i) factor and
resource conditions, including natural resources as well as labor; (ii) demand conditions;
(iii) related and supporting industries; (iv) firms’ strategies and industry structure. These
four elements must be considered together, as a part of an interactive whole. The focus
34
of Porter’s analysis is the innovative activity of firms, and the various economic factors
that may explain innovative activity and output in different countries.
Porter’s strictly economic approach to NIS is different from that of Nelson (1993). In
his book containing national case studies of the innovation systems of fifteen different
countries, Nelson divided them into three groups: large high-income, smaller high-
income and lower-income countries. In the concluding chapter, Nelson (1993, p.518)
argues that “it is inevitable that analysis of innovation in a country sometimes would get
drawn into discussion of labor markets, financial systems, monetary fiscal and trade
policies, and so on. One cannot draw a line neatly around those aspects of a nation’s
institutional structure that are concerned predominantly with innovation in a narrow
sense excluding everything else, and still tell a coherent story about innovation in a
broad sense”. The fundamental difficulty for these applied studies is precisely the one
stressed in the above quotation. It is rather difficult to define neatly which aspects to
include or to exclude from the study of an innovation system, as private and public
organizations interact within a complex institutional and economic structure. Empirical
research has then proceeded in a broad and open way, given that it is hard to find
universal and standard criteria to measure and compare the performance of innovation
systems in different countries.
The difficulties encountered by applied research suggest that a more structured
theoretical framework may be useful for the development of the approach. Closely
related to Nelson and Winter-like evolutionary tradition, theoretical research in the NIS
field started with the book National Systems of Innovation: Towards a Theory of
Innovation and Interactive Learning (Lundvall, 1992), which studied the
microeconomic foundations of the innovative process from an evolutionary and
systemic perspective. Innovation is a complex phenomenon of a dynamic, cumulative
35
and rather uncertain nature. It is important to consider such complexities not just from
the point of view of the single individual or firm, but rather by focussing on the
feedbacks and interactions between the various components of the system. Innovations
and learning are collective phenomena, they can hardly be understood without an
investigation of the complex interactions between heterogeneous economic agents.
A useful distinction in the systemic theory of innovation is that between ‘learning’ and
‘exploring’. Learning is a fundamental and ubiqutous characteristic of modern
knowledge-based economies. It is, first of all, the outgrowth of the productive process,
because individuals learn ‘by doing’ things (Arrow, 1962) and ‘by using’ technologies
(Rosenberg, 1982). But economic agents also learn ‘by interacting’ with other agents
(Lundvall, 1992). While learning is mainly an automatic and unaware consequence of
the working activities of individuals, ‘exploring’ denotes a deliberate and active effort
to search for new technical and organizational solutions, new products and processes.
Typically, it is the R&D system, and more generally the science and technology system,
to perform such an exploring activity in a systematic way. The main economic actors
involved in this process are R&D professional laboratories in private firms, and research
institutes and Universities in the public domain.
Private and public organizations in the science and technology system, however, are not
enough to define an innovation system, as there are important macro aspects that need
to be considered as well. In fact, any form of learning and exploring is “anchored in the
production structure and in the linkage pattern of the system of production” (Lundvall,
1992, p.17). So, an innovation system approach studies scientific and technological
activities within the whole productive system, in which learning continuously takes
place. In this respect, there are several factors that directly or indirectly affect the
innovative process: the education and training system in private firms and public
36
schools, the role of government in innovation and industrial policies, the general
macroeconomic conditions, and consumers’ tastes and competencies. This suggests an
interesting link in the NIS framework between the micro and the macroeconomic levels
of analysis. The innovative behavior of individuals and firms is affected by
macroeconomic and structural characteristics, and in turn the patterns of innovative
activity determine the evolution of those aggregate features over time.
This implicit interaction between the micro and the macro levels of analysis within an
innovation system is not only considered from a strictly economic point of view, but
also from an institutional one. Institutions are conceived in a broad way as “sets of
common habits, routines, established practices, rules or laws that regulate the relations
and interactions between individuals and groups” (Edquist and Johnson, 1997, p.46).
They differ from organizations, which are consciously created and have an explicit
purpose. Institutions can be thought of as ‘rules of the game’, while organizations are
regarded as ‘players’. The fundamental importance of institutions in the innovative
process is related to the fact that individuals are not rational agents, but follow habits of
thought. Economic agents do not properly choose what to do, as it is the case in
neoclassical microeconomic theory, but rather follow predetermined rules, routines,
habits of thought. When these habits are “common to the generality of men” (Veblen,
1919, p.239), they become ‘institutions’. It follows that any kind of human knowledge
and innovation is accumulated and stored through habits of thought, routines and
institutions. Cognitive, learning and innovative processes at the microeconomic level
are shaped by institutions and social structures; in turn, the aggregate level of analysis is
continuously affected by individuals’ innovative and learning processes, which tend to
modify the previously adopted patterns of behavior and habits of thought.
37
According to this view, institutions have an important function to perform in innovation
systems, in that they help economic agents to reduce the uncertainty and complexity of
the innovative process. Institutions have a stable character, which sustains and enables
individual learning and incremental innovations. At the same time, however, in periods
of rapid and radical technological change, there is a pressure for the old habits of
thought, routines and institutions to modify and to adapt to the emerging technological
paradigm. In such circumstances, some creative destruction of knowledge and
institutions is necessary, so that forgetting may be as important as learning (Johnson,
1992). Countries with a higher social and institutional adaptability are more likely to
enter quickly into a new techno-economic paradigm, and to have positive economic
performance for a prolonged period of time, as shown by the case of Japan (Freeman,
1987).
However, although the role of institutions and their relationship with the innovative
process is increasingly pointed out in the NIS approach, it must be recognised that the
systemic analysis of institutional transformations is still much less developed than the
corresponding analysis of technological change. The main focus of the research has so
far predominantly been on the role of innovative activities performed by private and
public organizations.
3.4.1 Discussion
In general system theory, a system is composed by its components and the relationships
between them (Edquist, 2005). What gives an evolutionary flavour to the NIS approach,
therefore, is not the use of system theory per se, but rather the way in which the
components and their relationships are represented. This is discussed as follows.
38
As regards the relations between micro actors and macro structure, the NIS framework
presents an explicit attempt to find a compromise between the two opposite views of
methodological individualism and methodological holism. In a non-reductionist fashion,
both levels of analysis, micro and macro, are studied in the framework. The innovative
activity of private and public organizations is affected and shaped by the production
structure, the macroeconomic conditions and the socio-institutional system of the
country; in turn, national patterns of innovation and productivity growth are determined
by the learning and searching activities of (micro) economic agents. The study of the
interactions between micro actors and social structure may constitute an important
contribution of the NIS framework to the development of evolutionary growth
theorizing. Important insights in this respect could come from a rediscovery of the work
of Veblen and old American institutionalism, based on the dynamic relationships
between ‘habits of thought’ and ‘institutions’ (Hodgson, 1993 and 1998). Such a non-
reductionist link between micro and macro levels of the analysis, however, has not been
made explicit yet in NIS studies, and needs to be further investigated in future research.
Focusing the attention on the microfoundations of the NIS framework, a fundamental
characteristic is the heterogeneity and variety of individuals and organizations
(McKelvey, 1997; Saviotti, 1997). If all economic agents had the same learning and
innovative capabilities, in fact, innovation could be studied by simply analysing the
behavior of a representative agent, and there would be no need of a systemic
perspective. But great complexities arise because learning and innovation are interactive
activities carried out by heterogeneous agents. NIS is therefore implicitly based on
population thinking.
Technical and organizational changes are the result of the complex interactions between
private and public organizations, consumers and users of new technologies,
39
macroeconomic structure and institutional framework. Innovation partly depends on
learning processes (an inherent consequence of the production process), and partly on
exploring activities (a deliberate effort to search for new technical solutions performed
by the science and technology system). Therefore, the NIS theory of innovation points
to the role played by both, radical and incremental innovations. It presents an
explanation of technical and organizational changes that combines the neo-
Schumpeterian focus on radical innovations with the evolutionary microfounded theory
based on learning and incremental changes.
Consequently, the dynamics of the economic process driven by innovation is in
principle consistent with both, a saltationist dynamics as in neo-Schumpeterian and
technology-gap approaches, and the more gradualist view typical of Nelson and Winter-
like evolutionary studies. Nevertheless, most empirical research has so far focused on
the static characteristics of the innovation system in a given period, not on its process of
change over time. The analytical-theoretical explanation of the process of evolution of
the system over time has not been made explicit yet in the NIS framework. The study of
‘complex evolving system’ could constitute, according to Metcalfe (2001), a way to
give a more precise evolutionary foundation to the NIS approach, and to investigate the
dynamic properties of evolving national systems in a more rigorous way.
What is clear is that, similarly to the other strands of research within evolutionary
economics, economic change is conceived as a non-deterministic and non-predictable
process. The fundamental and permanent source of uncertainty and unpredictability is
constituted by the complexities of the innovative process and by the nature of
interactive learning between heterogeneous individuals. Relatedly, as in all the other
evolutionary perspectives considered so far, it is also clear that the evolution of the
40
innovation system does not tend towards equilibrium, but it is assumed to be a never-
ending and always changing process.
4. New growth theory
This section shifts the focus to the other major paradigm in modern theorizing on
innovation and growth, namely new growth theory (NGT). The first part of the section
describes the main types of analytical models developed in the field in the last two
decades, while the second part points to the major strands of applied research related to
NGT. There exist already comprehensive overviews of this literature, in relation to both
NGT analytical models (Aghion and Howitt, 1998; Chui et al., 2002) and the related
applied work (Temple, 1999; Islam, 2003). The reader is referred to these previous
works for a more complete discussion of the technical issues involved, and for a more
in-depth assessment of the advances obtained in this field.15 The present section, on the
contrary, does not aim at providing a complete survey of NGT. The major purpose here
will be to prepare the ground for the discussion of the theoretical foundations of NGT,
which will be carried out in the final part of the section. The discussion will analyse the
theoretical foundations of NGT by discussing the six major questions that this paper
uses as framework of analysis, and this will make it possible to point out the great
differences existing between NGT and the evolutionary economic paradigm.
4.1 New growth models
New Growth Theory models originated in the second half of the 1980s to overcome the
problems left unresolved by the neoclassical model of economic growth (Solow, 1956).
Two were the main sources of criticism made to the Solovian view. First, under the
41
assumptions of constant returns to scale of each factor in the production function, and of
decreasing marginal productivity of capital over time, the Solow model predicted that
economic growth would have ceased in the long run. The only possible source of
permanent growth in the steady state was technological change, which was exogenous
and unexplained by the model. The latter was then unable to explain why GDP per
capita has been continuously growing in most industrialized countries since the
Industrial Revolution. Second, as technology was assumed to be a public good, freely
available to all countries, the model predicted that poor and rich countries would have
all converged to the same level and rate of growth of GDP per capita in the long run
(given population and saving rates). This prediction was in contrast with the empirical
evidence on the persistence of growth rates differences over long periods of time.
Taking these two problems into account, NGT models developed with the purpose of
explaining how technological change can generate sustained growth and persistent
differences between countries in the long run. The first generation of models was
pionereed by Romer (1986) and Lucas (1988). Their models suggest that technological
knowledge may be conceived as a non-rival good. This means that once new knowledge
is produced by a firm (or by an economic agent who is accumulating human capital) this
may benefit all the other firms as well. The public good characteristic of innovation
introduces an important externality in the economic system, and consequently it may
explain the existence of increasing returrns to scale in the aggregate production
function. Differently from the Solow model, an increase in inputs of production can
now have a permanent effect on the rate of growth of output, not only on its level. In the
new framework, then, a positive growth rate in the long run can be explained by
endogenous technical change, i.e. by the fact that there exist externalities associated to
the production of technological knowledge by economic agents.16 Moreover, an
42
important implication of these models is that endogenous technological change and
increasing returns in the aggregate production function may determine persistent
differences in economic growth rates between countries, and so tackle the second
question left unresolved by the Solovian model. The main problem associated with the
first generation of NGT models, however, was that the reason why economic agents
may decide to invest in the accumulation of knowledge and human capital was not made
explicit. If knowledge is a purely public good, in fact, where do individuals and firms
take the incentive to invest in the accumulation of these resources?
This question was considered by a second generation of models (Romer, 1990;
Grossman and Helpman, 1990). Still based on the idea that there are important
externalities associated to the public good features of knowledge, these models argue
that knowledge is an (at least partly) appropriable good, meaning that the fruits of
technical progress may be appropriated by the producer in the form of monopoly rents.
This idea is formalized by assuming that innovation is created by a separate research
sector, whose purpose is to create new blueprints for the production of intermediate
capital goods. Once a new blueprint is found, the producer firm can appropriate its
invention by patenting it, so that it becomes a monopolist in the production of the new
capital good. As a consequence, the assumption of perfect competition is released, as
the intermediate goods sector is characterized by monopolistic competition. These
models also assume that once a new capital good is produced, it adds to the older ones,
which are not instantaneously driven out of the market. Economic growth, then, takes
the form of an increasing variety of intermediate goods.
In a nutshell, the appropriability character of technological knowledge explains the
microeconomic incentive to invest in innovative activities, and its nonrival feature
explains aggregate increasing returns, endogenous growth and differences between
43
countries, as in the previous generation of models. The Romer (1990) and Grossman
and Helpman (1990) version of NGT, however, opens up an additional question: is it
appropriate to model innovation as a deterministic and certain outcome of the activity of
the research sector?
A third generation of NGT models, originating from the works of Aghion and Howitt
(1992) and Grossman and Helpman (1991), proposed an answer to this question by
pointing to the uncertain nature of innovative activity. Analytical models formalize the
uncertainty characterizing the innovative process by assuming that new blueprints are
found according to a Poisson stochastic process, whose parameter represents the
productivity in the research sector. As the parameter of the stochastic process is known,
it is possible to calculate an average arrival rate of innovation, and consequently an
average rate of growth of the economy. The second modification introduced by this
class of models is that they assume that each new blueprint makes the previous ones
instantaneously obsolete, so that the previous monopolists in the intermediate good
sector are driven out of the market as soon as an innovation is found. These models drop
the idea that there is an increasing variety of coexisting capital goods, and point out that
the process of technological competition and economic growth are characterized by
Schumpeterian ‘creative destruction’ (or ‘business stealing effect’): the monopoly
power associated to a new bluprint is only temporary, and once a new intermediate
capital good is introduced, the previous monopolists are driven out of the market.
Combining stochastic innovative activity, creative destruction and aggregate increasing
returns, these models predict that economic growth in the long run is a function of three
major factors: the amount of labour resources employed in the research sector, the
degree of market power in the intermediate capital goods sector, and the productivity in
the research sector.
44
An important empirical fact that these previous generations of endogeneous growth
models did not consider is the observation that innovations may have different sizes
with different impacts on the economy. The distinction between radical and incremental
innovations is an important one in evolutionary economics, and it was originally pointed
out in the neo-Schumpeterian long wave strand of research. How can this empirical fact
be represented in NGT models? How does the size of innovation matter for
endogeneous growth theory?
It is the most recent generation of NGT models (Bresnahan and Trajtenberg, 1995;
Helpman, 1998) to put forward an answer to these questions. These works emphasize
that innovations may have different sizes: they can be drastic (radical) or incremental.
Some drastic innovations, in particular, may have deep impacts on the process of
economic growth. According to Helpman (1998, p.13), “a drastic innovation qualifies as
a ‘general purpose technology’ if it has the potential for pervasive use in a wide range of
sectors in ways that drastically change their modes of operation”. From this definition, it
appears that the concept of ‘general purpose technologies’ (GPTs) is quite similar to
that of ‘technological paradigm’ (Dosi, 1982; Freeman, 1982), which we have
previously presented with reference to the neo-Schumpeterian long wave theory (see
section 3.1).
In this recent class of NGT models, once a new GPT arrives the radical innovation is
not immediately ready to be used in the final goods sector, but it needs to be
implemented in the form of a new intermediate capital good (incremental innovation).
The arrival of the latter is formalized as a Poisson-distributed random variable, as in the
previous Aghion and Howitt (1992) and Grossman and Helpman (1991) models.
After the introduction of the new GPT, labour resources are transferred to the research
sector in order to develop the new capital goods and to appropriate the relative
45
monopoly rents, in a phase in which the old technological paradigm still presents higher
productivity than the new one. At the aggregate level, this first phase results in a slump
of economic activity that may last for a few decades. Later on, once a certain number of
intermediate goods embodying the new GPT are found, the profitability of the new
methods of production turns out to be more evident to the firms in the final goods
sector, and the GPT becomes the new dominant technological paradigm. In this second
phase, the new GPT diffuses to the whole economy, and this may sustain the growth of
aggregate productivity for the following decades. This two-phase cycle of growth is
assumed to repeat over time, and in the long run such a cyclical trend tends towards the
steady state.
The stylized description presented above clearly indicates that GPT models propose a
formalization of the Schumpeterian theory of long waves. This is an interesting attempt,
but modeling exercises of this type should in the future address some major questions:
(i) Why does the GPT arrive at a given point in time? Is there any economic or socio-
institutional factor affecting the invention of new GPTs? (ii) Is there any economic or
socio-institutional factor affecting the rate of diffusion of the new technological
paradigm over time and in different countries? These questions suggest possible
avenues for future research in this field, and open up a possible ground for further
interactions between the evolutionary long wave theory and the GPT modeling tradition.
4.2 NGT applied research
The development of new growth models has attracted a great deal of interest of
empirically oriented scholars, and a huge amount of applied studies on cross-country
differences in economic growth has flourished in the last fifteen years. These works
typically take the form of cross-section econometric regressions where the growth of
46
GDP per capita over time is regressed on its level at the beginning of the period (a
proxy for the scope for catching up), and on a set of other structural and economic
characteristics, such as, for instance, countries’ accumulation of physical capital, and
levels of education and human capital.
The so-called convergence debate refers to the two different approaches and
interpretations that this type of study may lead to (Temple, 1999; Islam, 2003). One set
of econometric works derives growth regressions in the context of the Solow model
augmented with human capital (e.g. Mankiw, Romer and Weil, 1992). The convergence
property is then interpreted, in a neoclassical fashion, as a result of decreasing marginal
product of physical and human capital. Another stream of cross-country applied
research, on the contrary, includes additional variables in the specification, such as
political conditions, industrial structure, and so on, and interprets the conditional
convergence result (or lack of such) in a NGT framework, i.e. as an indication of
persistent growth rate differences across countries (e.g. Barro, 1991; Barro and Sala y
Martin, 1995). Here, conditional convergence does not depend on different rates of
accumulation of physical capital, but rather on the advantages that the international
diffusion of technologies may determine for catching up countries.
An important result in the growth regressions literature, though, is that convergence is
not a ubiquitous phenomenon, but it depends to a great extent on the countries included
in the sample under study. In a seminal paper, Baumol (1986) pointed out the existence
of three different convergence clubs in the world economy (OECD countries, centrally
planned economies, and less developed countries), and demonstrated that the patterns of
convergence greatly differ between these groups. Baumol’s idea has been refined in a
number of subsequent empirical studies (De Long, 1988; Baumol and Wolff, 1988;
Baumol et al., 1989; Baumol, 1994), which have all stressed the great variety of
47
macroeconomic performance in the world economy, and the striking differences
between the rapid growth of a restricted group of advanced economies and the static
patterns of less developed countries (Pritchett, 1997).
A well-known paper by Durlauf and Johnson (1995) developed this idea further, and
classified world countries into four groups according to their initial conditions (i.e.
initial levels of GDP per capita and of literacy rate). Their empirical study confirmed
the existence of different convergence clubs with markedly different characteristics and
growth behaviour. Instead of using the common cross-country regression framework,
Quah (1996a; 1996b; 1996c; 1997) studied the dynamics of the distribution of world
income, and found evidence of “emerging twin peaks”, i.e. the existence of polarization
and of increasing differences between rich and poor countries.
These applied works are all the more important in growth theory because, as observed
by Temple (1999, p. 150), it is “useful to draw distinctions between types of country.
[…] It is important to move away from characterizing the ‘average’ developing country,
and work towards a deeper understanding of differences.” It is thus increasingly felt in
growth theory “the need to acknowledge heterogeneity, and move away from techniques
based on ‘representative’ economies” (ibid., p. 150).
The important challenge that this strand of econometric studies presents for future
research in the field is twofold. First, it shows that a wide array of different econometric
methodologies may be used to shed new light on the issue of growth rate differences
and to complement the traditional cross-sectional approach: panel data methods (Islam,
1995), regression trees and other clustering techniques (Durlauf and Johnson, 1995),
and techniques for analysing the dynamics of the whole distribution of world income
(Quah, 1996a). Secondly, it suggests the need for further modelling efforts to provide a
theoretical explanation for the empirical findings of convergence clubs, polarization and
48
twin peaks. One possible direction would be to extend the multiple equilibria type of
models (e.g. Azariadis and Drazen, 1990; Galor, 1996), although the link between this
class of model and the NGT empirical work has not been made explicit yet.
One major criticism often made to the new growth empirics is that the various
econometric studies are tests of conditional convergence on a large sample of countries,
and not of innovation-driven endogenous growth for each single country belonging to
the sample (Fine, 2000). They are not estimation of the structural form of the analytical
NGT models, but rather tests of its reduced form. As Paul Romer (1994, p.11) argues,
“the convergence controversy […] represents a digression from the main story behind
endogenous growth theory”, and not a direct test of it.
Sharing this point of view, Jones (1995a; 1995b) shifts the attention to the time series
implications of new growth models, so originating a new class of empirical tests more
directly aimed at testing the predictions of the endogenous formalizations. As
mentioned above, NGT models predict a positive relationship between the amount of
labour resources employed in the research sector and the rate of economic growth. Jones
shows, however, that this prediction is in contrast with the empirical evidence, which
indicates that the steady rise of R&D intensity since the 1960s has not been associated
to increasing but to constant or decreasing economic growth rates.
This finding has recently inspired a new type of empirical tests of NGT models, which
focuses on the time series dimension of the growth process, and on the ‘scale effects’
implications of the analytical models (Jones, 1999; Greiner, Semmler and Gong, 2005).
This new direction of research is promising, although various methodological and data-
related problems may hamper the diffusion of this type of time series tests (Temple,
1999 and 2003; Islam, 2003). An interesting connection that would enrich even further
this line of empirical research could be the one between time series endogenous tests
49
and the recent class of GPT models, investigating the empirical relevance of the two-
phases long wave cycle generated by the emergence of a new GPT.
4.3 Discussion
NGT models and the related empirical works are based on the concept of aggregate
production function, meaning that the approach is macroeconomic but implicitly
microfounded. Economic agents are represented as rational maximizers of an
intertemporal profit or utility function. Moreover, they are conceived as fundamentally
homogenous. The use of the ‘representative agent’ metaphor makes it possible to study
the effects of the microeconomic behaviour on the macroeconomic level of analysis by
using relatively simple analytical and formal models. The approach is thus based on
typological thinking, which implies a less realistic description but a greater analytical
power of NGT models as compared to evolutionary studies.
When NGT models formalize the ideas of ‘variety of capital goods’ and ‘product
differentiation’ (Romer, 1990; Grossman and Helpman, 1990), these are rather different
from the way in which heterogeneity and variety are conceived in evolutionary
economics. In NGT, variety is a consequence of innovative and learning activities of
economic agents (reminding somewhat of the ‘increasing complexity’ of the growth
process described by Adam Smith and Herbert Spencer), not an essential precondition
of the process of economic growth (as in evolutionary economics).
In NGT models, innovation is a major source of economic growth. Technological
knowledge is formalized as a non-rival and partly appropriable economic good
produced by a separate research sector. Innovation may be explained by two kinds of
complementary mechanisms: learning by doing in relation to the accumulation of
knowledge and human capital (emphasized by Romer, 1986 and Lucas, 1988), and
50
‘exploring’ through the R&D activity of private firms (emphasized since the model of
Romer, 1990). More recently, GPT models enlarge the set of possible mechanisms
explaining technological innovation and productivity growth by assuming that radical
innovations may have particularly deep and pervasive impacts on the economy. At the
present stage of development of these models, however, the arrival of a new GPT is
exogenous, serendipitous, and not linked to any economic or socio-institutional factor.
A field for further modeling exercises would be to endogeneize the arrival of radical
innovations by following some of the insights coming from the neo-Schumpeterian long
wave literature (Freeman et al., 1982; Freeman and Louca, 2001; see section 3.1). An
interesting possibility, in particular, would be to model the ‘depression-trigger
hypothesis’ by linking the rate of arrival of radical innovations to economic factors such
as demand and profitability conditions in the downswing phase of the long wave cycle.
Learning by doing and exploring activities by the R&D sector determine a gradualist
type of dynamics, following which the economic system smoothly proceeds towards the
steady state. History, then, may be conceived as a uniform-speed transitional dynamics,
rather than an evolutionary process of transformation and qualitative change. Such a
gradualist view, however, may be modified in the future by the advances of the recent
generation of GPT models. In this case, innovations are mainly radical and may
determine saltationist dynamics and long run fluctuations. The temptative combination
of gradualist and saltationist dynamics in GPT models appears to be a novel element in
NGTs, and needs to be further explored in the future.
Another important theoretical feature of NGT models is the way in which uncertainty is
introduced in modeling exercises. These, in fact, represent innovation as an uncertain
outcome of R&D activities by assuming that its arrival rate follows a Poisson stochastic
process with given parameter. This formalization suggests that the process of growth is
51
not characterized by ‘strong’ and radical uncertainty as in evolutionary economics
(Dosi, 1982), but rather by ‘computable risk’. In fact, although a stochastic element
exists in the model, it is still possible to predict on average the rate of arrival of
innovations, and consequently that of economic growth. The recent class of GPT
models does not constitute an exception in this respect: the two-phase cycle repeats
mechanically over time, and no strong uncertainty is present in the succession of the
deterministic and predictable long waves. The way in which uncertainty is represented
in NGT models implies a stylized description of the growth process, but its advantage is
certainly the greater tractability and stronger analytical power of NGT models as
compared to evolutionary works. In an emerging class of models of economic
dynamics, namely chaos models, a deterministic system may, due to the high sensitivity
in initial conditions, generate radically uncertain outcomes (Boldrin and Woodford,
1990; George and Oxley, 1999). This type of models presents a peculiar combination of
neoclassical characteristics (the description and microfoundations of the deterministic
system) and evolutionary outcomes (uncertain and disequilibrium behaviour). This class
of models has not been applied yet to the study of innovation and growth, but its wider
use in future modeling and empirical exercises may possibly constitute a bridge between
the evolutionary and NGT research traditions.
Finally, the economic process represented by NGT, in formal models as well as in
empirical works, tends towards a steady state of balanced growth, which may differ
across countries. Differently from evolutionary economics, the impact of innovation on
economic growth is therefore analysed in a dynamic equilibrium setting. In GPT models
too, the economic fluctuations determined by the stochastic arrival of innovations do not
permanently deviate from the long run equilibrium trend. Temple (2003) has recently
argued, however, that the steady state metaphor should not be taken too literally by
52
growth researchers, as its major purpose is to provide an analytical tool for the
tractability of formal models, and not a prediction to be tested by empirical studies. Yet
another challenge for future research, then, would be to shift the focus from the long run
properties of the growth models towards the process of transitional dynamics, which is
all the more important in terms of welfare and policy implications (George, Oxley and
Carlaw, 2004).
Table 1. The theoretical foundations of evolutionary and new growth theories.
New Growth Theories Evolutionary theories
What is the main level
of aggregation?
Aggregate models based on neoclassical microfoundations (‘methodological individualism’)
Towards a co-evolution between micro and macro levels of analysis
(‘non-reductionism’)
Representative agent or
heterogeneous individuals?
Representative agent and
typological thinking
Heterogeneous agents and population thinking
What is the mechanism of
creation of innovation?
Learning by doing and ‘searching’ activity by the R&D sector;
Radical innovations and GPTs
Combination of various forms of learning with radical technical and
organizational innovations
What is the dynamics of the
growth process?
How is history conceived?
History is a uniform-speed transitional dynamics
Towards a combination of gradualist and saltationist dynamics:
history is a process of qualitative change and transformation
Is the growth process
deterministic or unpredictable?
‘Weak uncertainty’ (computable risk): stochastic
but predictable process
‘Strong’ uncertainty: non deterministic and unpredictable process
Towards equilibrium
or never ending?
Towards the steady state
Never ending and ever changing
53
5. Conclusions
The article has presented a critical survey of two major modern approaches to the study
of innovation and economic growth, evolutionary and new growth theories. The purpose
has been to discuss the often made claim that the two approaches, both inspired by
Schumpeter’s seminal works, are becoming more and more similar in terms of the
sources and mechanisms of the growth process on which they focus. According to this
argument, some kind of theoretical convergence between the two paradigms is taking
place.
The article has argued that a comprehensive comparison of these different growth
theories cannot simply be done by pointing to their common Schumpeterian features, as
it is frequently done, or by looking at the properties and results of modeling exercises.
The comparison needs to be made at a more general level of analysis, that is by
investigating the theoretical foundations of the different approaches. By theoretical
foundations we mean the theoretical characteristics that may be considered as the main
building blocks of each growth paradigm.
Following this idea, section 2 has presented the six theoretical questions that we have
used as a framework to compare the two approaches. It has defined the main concepts
used in the survey, and it has briefly pointed to the origin of each concept in the history
of economic thought. Section 3 has analysed the basic foundations of different streams
of modern evolutionary economics, namely the neo-Schumpeterian long wave theory,
the technology-gap approach, Nelson and Winter-like evolutionary theorizing, and the
national innovation systems framework. The section has shown that these recent streams
of evolutionary economics share the same theoretical foundations, so that they can be
regarded as different strands of research within the same (broadly defined) evolutionary
54
paradigm. However, the discussion has also pointed to some existing tensions and to the
great challenges ahead for the evolutionary economic paradigm.
Section 4 has turned the attention to new growth theory, and it has briefly discussed the
main generations of analytical models, as well as the major developments in the applied
tradition. The section has shown that the theoretical foundations of new growth theory
greatly differ from those of the evolutionary approach. The main differences between
the two growth paradigms can be summarized as follows (see table 1).
(1) The aggregate properties of new growth models are derived from the analysis of the
behaviour of rational economic agents, and the related cross-country econometric work
is set up in a production function framework. Both of them are thus implicitly based on
methodological individualism. On the contrary, evolutionary studies point to the
theoretical advantages of a non-reductionist theory where the micro and macro levels of
analysis co-evolve and interact with each other. This attempt is often called for, but it is
nonetheless difficult to make operational, and the different strands of evolutionary
research have not yet reached a clear and common position in this respect.
(2) ‘Typological thinking’ and ‘representative agent’ are conceived as useful principles
in new growth theory, as they increase the analytical tractability of formal models, thus
strengthening their conceptual power. In NGTs, the notion of heterogeneity is not an
essential intrinsic characteristic of individuals, firms, sectors and countries, but rather a
consequence of the productive process, close in spirit to the metaphor of increasing
complexity associated with the growth process described in the past by Adam Smith and
Herbert Spencer. In evolutionary economics, on the other hand, heterogeneity of
economic agents, routines and habits of thought assumes a fundamental role in the
construction of the theory, which is then close to a Veblerian type of population
55
thinking. The latter increases the realism of the description of the growth process, but
presents formidable challenges for modeling exercises.
(3) Although innovation is the main source of economic growth in both equilibrium and
evolutionary views, the underlying concept of knowledge is rather different. In new
growth theory, knowledge is conceived as a non-rival and partly appropriable economic
good. Evolutionary theories, though, point out that knowledge is a more complex entity,
which cannot be analysed in purely economic terms. According to evolutionary scholars
(Nelson and Winter, 1982), knowledge is often tacit, and not always codified and
codifiable. It is embodied in the routines of individuals and organizations, and not
stored in a book of blueprints. It is interactive, collective and systemic, and not simply
the result of individual learning. It tends to be highly dependent and strongly rooted in a
given organizational and institutional context, and not separable from it. On the whole,
the mechanisms of creation of innovation look similar in evolutionary and new growth
theories, but the conceptual foundation behind them is rather different.
(4) NGT models conceive history as a uniform-speed transitional dynamics towards the
steady state, not as a process of qualitative change and transformation. Evolutionary
economics, on the contrary, searches for a combination of saltationist and gradualist
dynamics, and stresses the role of qualitative change and permanent transformation of
the growth process. In both paradigms, however, it is difficult to combine gradualist and
saltationist features in a single theoretical framework, and this presents interesting
challenges for future research.
(5) The new growth world is characterized by ‘weak’ uncertainty and computable risk,
as implied by the use of random variables to formalize the arrival of innovation in the
analytical models. The growth process is hence stochastic but predictable. On the other
hand, the evolutionary growth process unfolds in an economic environment marked by
56
‘strong’ uncertainty and unpredictability. This is clearly argued by a large set of
appreciative and non-formal type of studies, while evolutionary modeling exercises do
not significantly differ from new growth models in this particular respect. The tension
between appreciative and formal types of evolutionary studies poses a crucial challenge
for future developments of the evolutionary paradigm.
(6) In NGTs, economic growth tends towards the steady state in the long run. The
steady state metaphor, in this context, should be interpreted as a useful tool that
increases the tractability and analytical power of formal models, rather than a prediction
to be confronted with empirical evidence. The growth path described by evolutionary
theories, on the contrary, is an ever changing and never ending process of change and
transformation, much closer in spirit to the disequilibrium economic world theorized in
the past by Veblen and Schumpeter.
In a nutshell, new growth theory combines ideas from classical authors such as Smith
and Schumpeter, and interpret them in a dynamic equilibrium framework, where
rational choices of economic agents lead to steady state outcomes in a stochastic way.
Evolutionary economics draws inspiration from various classical authors, such as Marx,
Veblen and Schumpeter, and interpret their insights in an evolutionary disequilibrium
context, where interactions among routine-guided and boundedly rational heterogeneous
agents determine an unpredictable and endless process of qualitative change and
transformation. The former paradigm points to the advantages that formal modelling
may lead to in terms of increased analytical simplicity and greater power of
generalization. The latter stresses the new insights that a more realistic description of
the growth process makes it possible to obtain.
57
On this ground, the often-claimed convergence between evolutionary and new growth
theories cannot be simply justified in terms of their common Schumpeterian features.
The analysis carried out in this paper leads to the conclusion that evolutionary and new
growth theories greatly differ with respect to all of their theoretical foundations. No
theoretical convergence between the two paradigms is taking place.
This finding should be welcomed by both evolutionary and new growth scholars, as it is
not theoretical convergence that determines advances in growth theory, but rather the
continuous process of interaction and give-and-take between the two paradigms.
Although no theoretical convergence is taking place, in fact, there exists an intense
exchange of ideas and a fruitful interaction between the two approaches. On the one
hand, evolutionary economics greatly benefits from the development of NGTs. The
unsatisfaction with the stylized and formal type of analysis of the development process
offered by endogenous growth models has proved to be a fundamental motivation to
induce evolutionary economists to provide more realistic descriptions and to search for
new empirical insights.
On the other hand, new growth theories benefit from the development of evolutionary
economics, as the latter provides new insights on the complexities associated with the
innovative process and its impacts on economic performance. The re-interpretation of
some of these evolutionary insights in a dynamic equilibrium framework has in fact led
to the refinement of NGT models and to new empirical applications. Three specific
examples may illustrate this point. The first refers to the evolutionary strand of long
wave theory flourished in the 1980s (section 3.1). This type of historical and descriptive
research has later been formalized by the recent class of GPTs models, where the
evolutionary insights on radical innovations, technological paradigms and
Schumpeterian long waves have been re-interpreted in an endogenous growth
58
framework. The second example relates to the technology-gap approach (section 3.2).
These type of empirical studies, flourished during the 1980s, were originally quite close
to an evolutionary and disequilibrium interpretation of the growth process. But a later
strand of econometric work in NGT applied a similar idea on the relevance of
innovation and the international diffusion of new technologies, and interpreted it in the
context of a micro-founded dynamic equilibrium setting. Finally, a third type of
interaction between the two paradigms refers to the idea of variety of macroeconomic
behaviour. While this has been a major point motivating evolutionary research since its
outset (e.g. in the literature on national systems of innovation, see section 3.4),
mainstream growth theory did not initially acknowledge this as a major point for
building up analytical models and untertaking empirical research. In the last decade,
however, NGT has increasingly focused on the great variety of growth behaviour in the
world economy, and investigated the existence of different convergence clubs through
multiple equilibria models as well as a wide array of non-parametric econometric
techniques.
In all these examples, the insights provided by evolutionary research have proved to be
a crucial motivation to develop successive waves of new growth models, where the
latter have re-interpreted the evolutionary insights in a mainstream dynamic equilibrium
framework based on neoclassical microfoundations. The outcomes of these NGT
models reproduce the same stylized facts pointed out by evolutionary studies, but,
admittedly, the theoretical structure underpinning them is fundamentally different from
the conceptual framework originally proposed by evolutionary theories. So, the cases
mentioned above do not represent examples of theoretical convergence between the two
paradigms, but they rather indicate the existence of interactions between radically
different economic worlds.
59
The interactions between these alternative paradigms have been quite important for the
development of the field in the last two decades, and they will go on playing a relevant
role in the future. The crucial point is that such fruitful exchange of ideas between
evolutionary and new growth theories takes place precisely because the two approaches
are so different. Therefore, it is the inherent difference between the two that stimulates
advances in growth theory, not their convergence to a common paradigm. The day in
which different paradigms will have converged to a single framework, growth theory
will cease to be such a dynamic and fascinating field of research. This day is still distant
in the future.
60
Acknowledgements I wish to thank Michele Di Maio, Jan Fagerberg, Arne Fevolden, Jarle Hildrum, Per Botolf Maurseth, Bart Verspagen and an anonymous referee of this Journal for their helpful comments and suggestions on a previous draft of the paper. The usual disclaimers apply.
Notes
1. Following Witt (1991), Nelson (1995), and Fagerberg (2003), we will use the label ‘evolutionary economics’ to indicate the whole set of approaches that will be discussed in section 3. Section 3.3, however, will focus on the more narrow set of evolutionary approaches directly linked to Nelson and Winter (1982)’s influential theory of economic change, that we will label ‘Nelson and Winter-like evolutionary theorizing’.
2. There are some other important disequilibrium views (such as the Austrian School, Post-Keynesian, and Institutional economics) that are indirectly related to the development of modern evolutionary economics, but lay outside of the scope of this survey, so that they will not be considered further.
3. A more detailed discussion of the relevance of different levels of aggregation in economic theory can be found in Hodgson (1993, ch.15).
4. It is well known that Schumpeter was the first to use the expression ‘methodological individualism’. 5. See Swedberg and Granovetter (1992). 6. See also Hodgson (1998). 7. An extended discussion of the concept of ‘population thinking’ can be found in Andersen (1994) and
Hodgson (1993). 8. Freeman (1994) and Dosi (1997) present critical surveys of the different mechanisms of technical
change in economic theory. 9. A critical discussion of the role of technical change in the theory of Karl Marx can be found in Elster
(1983) and Hodgson (1993). 10. See Hodgson (1993, ch.14) for a critical discussion of this issue. The brief characterization of a
mechanistic, deterministic and predictable economic process pointed out here is admittedly simplistic. There exists a class of models of economic dynamics, so-called chaos models (Boldrin and Woodford, 1990; George and Oxley, 1999), where a deterministic system, due to the high sensitivity in initial conditions, may lead to stochastic behaviour and uncertain outcomes. However, chaos models have not been widely applied yet to the study of innovation and growth, and a discussion of them goes therefore beyond the scope of this survey.
11. The historically oriented literature on catching-up and growth has been recently surveyed by Fagerberg and Godinho (2005).
12. Pavitt (2005) has recently considered some of the most important contributions in this now huge literature.
13. An overview of the main findings of the recent studies of sectoral systems of innovation can be found in Malerba (2005).
14. For a previous discussion of evolutionary models of economic growth, with special emphasis on diffusion models, see Sarkar (1998).
15. Several comprehensive surveys related to NGT have recently been presented in this Journal, in relation to different aspects of growth theory, such as the role of trade for the growth process (Lewer and Van den Berg, 2003), the effects of inflation (Temple, 2000; Gillman and Keyak; 2005) and of financial liberalization (Auerbach and Siddiki, 2004), and the role of fiscal policies (Zagler and Durnecker, 2003). Measurement and empirical issues have also been discussed, particularly in relation to different measures of human capital (Wobmann, 2003; Gibson and Oxley, 2003) and of TFP (Carlaw and Lipsey, 2003). For a detailed discussion of these contributions, see George, Oxley and Carlaw (2004).
16. New growth models are in fact also referred to as endogenous growth models, due to the endogenous nature of technological change. All the models reviewed in this section share this characteristic, as they all focus on innovation as the main engine of growth. However, there exist other classes of endogenous growth models that emphasize other sources of economic dynamics than technological change (see previous note). In this type of models, which we do not consider in this survey, it is the long run growth rate, rather than innovation, that constitutes the endogenous feature of the formalization.
61
References
Abramovitz, M. (1986) Catching-up, forging ahead and falling behind. Journal of
Economic History 46, 385-406. ⎯ (1994) The origins of the postwar catch-up and convergence boom. In Fagerberg, J., Verspagen, B. and von Tunzelmann, N. (eds.) The Dynamics of Technology, Trade and
Growth. Aldershot: Edward Elgar. Aghion, P. and Howitt, P. (1992) A model of growth through creative destruction. Econometrica 60, 323-351. ⎯ (1998) Endogenous Growth Theory. Cambridge, MA: MIT Press. Alchian, A. A. (1951) Uncertainty, Evolution, and Economic Theory. Journal of Political
Economy 58, 211-222. Andersen, E. S. (1994) Evolutionary Economics, Post-Schumpeterian Contributions, London: Pinter. Archibugi, D. and Michie, J. (1998) Technical change, growth and trade: new departures in institutional economics. Journal of Economic Surveys 12 (3), 313-332. Arrow, K. (1962) The economic implications of learning by doing. Review of Economic
Studies 29, 155-173. Arthur, W. B. (1994) Increasing Returns and Path Dependence in the Economy. Ann Arbor, MI: University of Michigan Press. Auerbach, P. and Siddiki, J.U. (2004) Financial liberalisation and economic development: an assessment. Journal of Economic Surveys 18(3), 231-265. Azariadis, C., Drazen, A. (1990) Threshold externalities in economic development. Quarterly Journal of Economics 1990, 501-526. Barro, R. (1991) Economic growth in a cross-section of countries. Quarterly Journal of
Economics 106, 407-443. Barro, R. and Sala Y Martin, X. (1995) Economic Growth. Boston: McGraw-Hill. Baumol, W. J. (1986) Productivity growth, convergence and welfare: what the long-run data show. American Economic Review 76 (5), 1072-1085. ⎯ (1994) Multivariate growth patterns: contagion and common forces as possible sources of convergence. In: W. J. Baumol, R. R. Nelson, E. N. Wolff, (eds.) Convergence of Productivity: Cross-National Studies and Historical Evidence, Oxford. Baumol, W. J., Wolff, E. N. (1988) Productivity growth, convergence and welfare: reply. American Economic Review 78 (5), 1155-1159.
62
Baumol, W. J., Batey Blackman, S. A., Wolff, E. N. (1989) Productivity and American
Leadership: the Long View. Cambridge, Mass.: MIT Press. Bresnahan, T. and Trajtenberg, M. (1995) General purpose technologies: ‘engines of growth’?. Journal of Econometrics 65, 83-108. Boldrin, M. and Woodford, M. (1990) Equilibrium models displaying endogenous fluctuations and chaos. Journal of Monetary Economics, 25, 189-222. Cappelen A., Castellacci, F., Fagerberg J. and Verspagen, B. (2003) The impact of regional support on growth and convergence in the European Union. Journal of
Common Market Studies 41 (4), 621-644. Carlaw, K.I. and Lipsey, R.G. (2003) Productivity, technology and economic growth: what is the relationship?. Journal of Economic Surveys 17(3), 457-495. Chiaromonte, F. and Dosi, G. (1993) Heterogeneity, competition, and macroeconomic dynamics. Structural Change and Economic Dynamics 4, 39-63. Chui, M., Levine, P., Murshed, S.M. and Pearlman, J. (2002) North-South models of growth and trade. Journal of Economic Surveys 16(2), 123-165. Conlisk, J. (1989) An aggregate model of technical change. Quarterly Journal of
Economics 104, 787-821. Cornwall, J. (1977) Modern Capitalism, its Growth and Transformations. London: Martin Robertson. David, P. (1985) Clio and the economics of QWERTY. American Economic Review 75, 332-337. De Long, J.B. (1988) Productivity growth, convergence and welfare: comment. American Economic Review 78 (5), 1138-1154. Dosi, G. (1982) Technological paradigms and technological trajectories. Research
Policy 11, 147-162. ⎯ (1997) Opportunities, incentives and the collective patterns of technological change. Economic Journal 107, 1530-1547
Dosi, G. Pavitt, K. and Soete, L. (1990) The Economics of Technical Change and International Trade, London: Harvester Wheatsheaf. Dosi, G. and Fabiani, S. (1994) Convergence and divergence in the long-term growth of open economies. In G. Silverberg and L. Soete (eds.) The Economics of Growth and
Technical Change: Technologies, Nations, Agents. Aldershot: Edward Elgar. Dosi, G., Fabiani, S., Aversi, R. and Meacci, M. (1994) The dynamics of international differentiation: a multy-country evolutionary model. Industrial and Corporate Change 3, 225-241.
63
Dosi, G. and Winter, S. (2000) Interpreting economic change: evolution, structures and games. LEM Working Paper 2000/08. Durlauf, S. N., Johnson, P. A. (1995) Multiple regimes and cross-country growth behaviour. Journal of Applied Econometrics 10, 365-384. Edquist, C. (2005) National systems of innovation. In J. Fagerberg, D. C. Mowery and R. R. Nelson (Eds.) The Oxford Handbook of Innovation, Oxford: Oxford University Press. Edquist, C. and Johnson, B. (1997) Institutions and organizations in systems of innovations. In C. Edquist (ed.) Systems of Innovation: Technologies, Institutions and
Organisations. London and Washington: Pinter. Elster, J. (1983) Explaining Technical Change. Cambridge, MA: Cambridge University Press. Fagerberg, J. (1987) A technology gap approach to why growth rates differ. Research
Policy 16, 87-99. ⎯ (1988) International competitiveness. Economic Journal 98, 355-374. ⎯ (1994) Technology and International differences in growth rates. Journal of
Economic Literature 32, 1147-1175. ⎯ (2003) Schumpeter and the revival of evolutionary economics: An appraisal of the literature. Journal of Evolutionary Economics 13, 125-159. Fagerberg, J. and Verspagen, B. (1996) Heading for Divergence? Regional Growth in Europe Reconsidered. Journal of Common Market Studies 34, 431–448. ⎯ (2002) Technology-gaps, innovation-diffusion and transformations: an evolutionary interpretation. Research Policy 31, 1291-1304. Fagerberg, J. and Godinho, M. M. (2005) Innovation and Catching-up. In J. Fagerberg, D. C. Mowery and R. R. Nelson (Eds.) The Oxford Handbook of Innovation, Oxford: Oxford University Press. Fagiolo, G. and Dosi, G. (2003) Exploitation, exploration and innovation in a model of endogenous growth with locally interacting agents. Structural Change and Economic
Dynamics 14 (3), 237-273. Fine, B. (2000) Endogenous growth theory: a critical assessment. Cambridge Journal of
Economics 24, 245-265. Freeman, C. (1983) Long Waves in the World Economy. Butterworth, Kent. ⎯ (1984) Prometheus unbound. Futures October 1984, 490-500.
64
⎯ (1987) Technology Policy and Economic Performance: Lessons from Japan. London: Pinter. ⎯ (1994) The economics of technical change. Cambridge Journal of Economics 18, 463-514.
Freeman, C., Clark, J. and Soete, L. (1982) Unemployment and Technical Innovation. London: Pinter. Freeman, C. and Louça F. (2001) As Time Goes by: From the Industrial Revolutions to the Information Revolution. Oxford: Oxford University Press. Galor, O. (1996) Convergence? Inferences from theoretical models. Economic Journal 106, 1056-1069. George, D. and Oxley, L. (1999) Robustness and local linearisation in economic models. Journal of Economic Surveys 13(5), 529-550. George, D., Oxley, L. and Carlaw, K. (2004) Surveys in Economic Growth: Theory and Empirics. Blackwell Publishing. Gerschenkron, A. (1962) Economic Backwardness in Historical Perspective. Cambridge, MA: Harvard University Press. Gibson, T.L.J. and Oxley, L. (2003) Cost- and income-based measures of human capital. Journal of Economic Surveys 17(3), 271-307. Gillmann, M. and Kejak, M. (2005) Contrasting models of the effect of inflation on growth. Journal of Economic Surveys 19(1), 113-136. Greiner, A., Semmler, W., Gong, G. (2005) The Forces of Economic Growth: A Time
Series Perspective. Princeton: Princeton University Press. Grossman, G.M. and Helpman, E. (1990) Comparative advantages and long run growth. American Economic Review. 80, 796-815. ⎯ (1991) Quality ladders in the theory of growth. Review of Economic Studies 58, 86-91. Habakkuk, H.J. (1962) American and British technology in the Nineteenth Century. Cambridge, MA: Cambridge University Press. Heertje, A. (1993) Neo-Schumpeterians and economic theory. In Magnusson (ed.) Evolutionary Approaches to Economic Theory. Dordrecht: Kluwer. Helpman, E., ed. (1998) General Purpose Technologies and Economic Growth. Cambridge, MA: MIT Press. Hodgson, G. (1993) Economics and Evolution: Bringing Life Back into Economics. Cambridge, MA: The MIT Press.
65
⎯ (1997) The evolutionary and non-Darwinian economics of Joseph Schumpeter. Journal of Evolutionary Economics 7(2), 131-145. ⎯ (1998) The Approach of institutional economics. Journal of Economic Literature 36, 166-192. Islam, N. (1995) Growth empirics: a panel data approach. Quarterly Journal of
Economics 110, 1127-1170. ⎯ (2003) What have we learnt from the convergence debate?. Journal of Economic
Surveys 17(3), 309-362. Iwai, K. (1984) Schumpeterian dynamics. An evolutionary model of innovation and imitation. Journal of Economic Behavior and Organization 5, 159-190.
Johnson, B. (1992) Institutional learning. In Lundvall, B-Å. (ed.) National Systems of
Innovation: Towards a theory of Innovation and Interactive Learning. London: Pinter Publishers. Jones, C. (1995a) R&D based models of economic growth. Journal of Political
Economy 103, 759-784. ⎯ (1995b) Time series tests of endogenous growth models. Quarterly Journal of
Economics 110, 495-525. ⎯ (1999) Growth: with or without scale effects?. American Economic Review 89, 139-144.
Kleinknecht, A. (1981) Observations on the Schumpeterian swarming of innovations. Futures 13, 293-307. Knight, F. H. (1921) Risk, Uncertainty and Profit. Boston, MA, Houghton Mifflin. Kuznets, S. (1940) Schumpeter’s business cycles. American Economic Review 30, 257-271.
Landes, D. (1969) The Unbound Prometeus. Technological Change and Industrial Development in Western Europe from 1750 to the Present. Cambridge, Cambridge University Press. Levine, R. and Renelt, D. (1992) A sensitivity analysis of cross-country growth regressions. American Economic Review 82(4), 942-963. Lewer, J.J., and Van den Berg, H. (2003) How large is international trade’s effect on economic growth?. Journal of Economic Surveys 17(3), 363-396. Lucas, R. (1988) On the mechanics of economic development. Journal of Monetary
Economics 22, 3-42.
66
Lundvall, B-Å. (1992) National Systems of Innovation: Towards a theory of Innovation and Interactive Learning. London: Pinter Publishers. Malerba, F. (2002) Sectoral systems of innovation and production. Research Policy 31 (2), 247-264 . ⎯ (2005) How innovation differ across sectors and industries. In J. Fagerberg, D. C. Mowery and R. R. Nelson (Eds.) The Oxford Handbook of Innovation, Oxford: Oxford University Press. Mankiw, N., Romer, D., and Weil, D. (1992) A contribution to the empirics of economic growth. Quarterly Journal of Economics 107, 407-437. McKelvey, M. (1997) Using evolutionary theory to define systems of innovation. In Edquist, C. (ed) Systems of Innovation: Technologies, Institutions and Organisations.
London and Washington: Pinter. Mensch, G. (1979) Stalemate in Technology: Innovations Overcome the Depression. New York: Ballinger. Metcalfe, S. (2001) Institutions and economic progress. Industrial and Corporate
Change 10, 561-586. Nelson, R.R. (1993) National Innovation Systems: A Comparative Analysis. New York and Oxford: Oxford University Press. ⎯ (1995) Recent evolutionary theorizing about economic change. Journal of Economic
Literature 33, 48-90 Nelson, R. and Winter, S. (1977) In search of a useful theory of innovation. Research
Policy 6, 36-76. ⎯ (1982) An Evolutionary Theory of Economic Change. Cambridge: The Belknap Press of Harvard University Press. ⎯ (2002) Evolutionary theorizing in economics. Journal of Economic Perspectives 16 (2), 23-46. Pavitt, K. (1984) Sectoral patterns of technical change: towards a taxonomy and a theory. Research Policy 13, 343-373. ⎯ (2005) Innovation processes. In J. Fagerberg, D. C. Mowery and R. R. Nelson (Eds.) The Oxford Handbook of Innovation, Oxford: Oxford University Press.
Perez, C. (1983) Structural change and assimilation of new technologies in the economic and social systems. Futures October 1983, 357-375. ⎯ (1985) Microelectronics, long waves and world structural change: new perspectives for developing countries. World Development 13, 441-463
67
Porter, M. (1990) The Competitive Advantage of Nations. London: Macmillan. Pritchett, L. (1997) Divergence, big time. Journal of Economic Perspectives 11 (3), 3-17. Quah, D. T. (1996a) Twin peaks: growth and convergence in models of distribution dynamics. Economic Journal 106, 1045-1055.
⎯ (1996b) Convergence empirics across economies with (some) capital mobility. Journal of Economic Growth 1 (1), 95-124.
⎯ (1996c) Empirics for economic growth and convergence. European Economic
Review 40, 1353-1375.
⎯ (1997) Empirics for growth and distribution: stratification, polarization, and convergence clubs. Journal of Economic Growth 2, 27-59. Romer, P. (1986) Increasing returns and long-run growth. Journal of Political Economy, 94, 1002-1037. ⎯ (1987) Crazy explanations for the productivity slowdown. In Fischer, S. (ed.) NBER
Macroeconomics annual. Cambridge, MIT Press. ⎯ (1990) Endogenous technological change. Journal of Political Economy 98, 71-102. ⎯ (1994) The origins of endogenous growth. Journal of Economic Perspectives 8(1), 3-22. Rosenberg, N. (1982) Inside the Black Box: Technology and Economics. Cambridge: Cambridge University Press. Ruttan, V. (1997) Induced innovation, evolutionary theory and path dependence: sources of technical change. Economic Journal 107, 1520-1529. Sarkar, J. (1998) Technological diffusion: alternative theories and historical evidence. Journal of Economic Survey 12 (2), 131-176. Saviotti, P. (1997) Innovation systems and evolutionary theories. In Edquist, C. (ed) Systems of Innovation: Technologies, Institutions and Organisations. London and Washington: Pinter. Schumpeter, J. (1934) The Theory of Economic Development. Cambridge: Harvard University Press. ⎯ (1939) Business Cycles. Philadelphia: Porcupine Press. ⎯ (1943) Capitalism, Socialism and Democracy. New York: Harper. ⎯ (1954) History of Economic Analysis London: Routledge.
68
Silverberg, G., Dosi, G. and Orsenigo, L. (1988) Innovation, Diversity and Diffusion: A Self-Organisation Model. Economic Journal 98, 1032-54. Silverberg, G. and Lehnert, D. (1994) Growth fluctuations in an evolutionary model of creative destruction. In Silverberg, G. and Soete, L. (ed) The Economics of Growth and
Technical Change. Technologies, Nations, Agents. Aldershot: Edward Elgar. Silverberg, G. and Verspagen, B. (1994a) Learning, Innovation and Economic Growth: A Long-Run Model of Industrial Dynamics. Industrial and Corporate Change 3, 199-223. ⎯ (1994b) Collective Learning, Innovation and Growth in a Boundedly Rational, Evolutionary World. Journal of Evolutionary Economics 4, 207-226. ⎯ (1995) An Evolutionary Model of Long Term Cyclical Variations of Catching Up and Falling Behind. Journal of Evolutionary Economics 5, 209-227. ⎯ (1996) From the Artificial to the Endogenous: Modelling Evolutionary Adaptation and Economic Growth. In E. Helmstädter and M. Perlman (eds.) Behavorial Norms,
Technological Progress and Economic Dynamics: Studies in Schumpeterian Economics, Ann Arbor, MI: University of Michigan Press. Solow, R. (1956) A contribution to the theory of economic growth Quarterly Journal of
Economics 70, 65-94. Swedberg, R. and Granovetter, M. (1992) Introduction. In Granovetter, M. and Swedberg, R. (ed.), The Sociology of Economic Life. Westview Press. Temple, J. (1999) The new growth evidence. Journal of Economic Literature 37, 112-156. ⎯ (2000) Inflation and growth: stories short and tall. Journal of Economic Surveys 14(4), 395-426. ⎯ (2003) The long-run implications of growth theories. Journal of Economic Surveys 17(3), 497-510. Van Duijn, J.J. (1983) The Long Wave in Economic Life. London: Allen and Unwin. Veblen, T. (1899) The Theory of the Leisure Class: An Economic Study of Institutions. New York: Macmillan. ⎯ (1915) Imperial Germany and the Industrial Revolution. New York: Macmillan. ⎯ (1919) The Place of Science in Modern Civilisation and Other Essays. New York: Huebsch. Verspagen, B. (1991) A new empirical approach to catching up or falling behind. Structural Change and Economic Dynamics 2, 359-380.
69
⎯ (1993) Uneven Growth Between Interdependent Economies: Evolutionary View on Technology-gaps, Trade and Growth. Aldershot: Avebury. ⎯ (2005) Innovation and Economic Growth. In J. Fagerberg, D. C. Mowery and R. R. Nelson (Eds.) The Oxford Handbook of Innovation, Oxford: Oxford University Press. Winter, S. (1964) Economic ‘natural selection’ and the theory of the firm. Yale
Economic Essays 4, 225-272. ⎯ (1971) Satisficing, selection and the innovating remnant. Quarterly Journal of
Economics 85, 237-261. ⎯ (1984) Schumpeterian Competition in Alternative Technological Regimes. Journal of
Economic Behavior and Organization 5, 137-158.
Witt, U. (1991) Reflections on the present state of evolutionary economic theory. In G. Hodgson and E. Screpanti (eds.), Rethinking Economics: Markets, Technology and
Economic Evolution. Aldershot: Edward Elgar. Wobmann, L. (2003) Specifying human capital. Journal of Economic Surveys 17(3), 239-270. Zagler, M. and Durnecker, G. (2003) Fiscal policy and economic growth. Journal of
Economic Surveys 17(3), 397-418.
70